How to Read IVF Success Rates (Without Being Misled)

A plain-language guide to reading the one number a fertility clinic wants you to see, and the four hidden choices behind it.

In 2022, fertility clinics in the United States reported 435,426 ART cycles on 251,542 patients across 457 clinics, and those cycles produced 98,289 babies - CDC ART Surveillance. Roughly one in every 37 US babies that year was conceived through IVF or another form of assisted reproduction - USAFacts. It is one of the most heavily measured medical procedures in the country, with a federal law requiring every clinic to report its outcomes. And yet the single "success rate" a clinic puts on its homepage can be close to meaningless if you do not know how it was built.

That is not an accusation of fraud. It is a statement about arithmetic. A success rate is a fraction, and a fraction has a top (the numerator) and a bottom (the denominator). Change what you count on the top, change what you count on the bottom, change which patients you include, and the same clinical reality can be truthfully described as 19% or 43% or 69%. Every one of those numbers can be honest. Only one of them answers the question you are actually asking, which is: what is the realistic chance that a person like me, starting treatment at a clinic like this, ends up holding a baby.

This guide teaches you to read that fraction. It walks through the four denominators clinics use, the difference between a pregnancy and a live birth, why age is the single most important filter, how the mix of patients a clinic accepts moves its numbers without any change in quality, and the specific ways a headline figure gets engineered to look larger. It is not medical advice, and it will not tell you which clinic to choose. It will make you much harder to mislead. Every figure below links to the public record it came from, which is the same standard we hold ourselves to on IVFcost.co: show the number, name the source, and never dress up an estimate as a fact.

Contents

1. Why one success-rate number is never enough
2. The four denominators, and why they disagree
3. The numerator problem: a positive test is not a baby
4. Age is the filter that changes everything
5. Patient mix: how who a clinic treats moves its number
6. The comparison trap: why the regulators say do not compare clinics
7. Donor eggs quietly swap the denominator
8. Cycle counts, cancellations, and batching
9. Reading a SART or CDC report line by line
10. The reporting lag: what "2022 data" actually means
11. The marketing playbook: how a number is made to look bigger
12. A practical checklist before you trust any rate

1. Why one success-rate number is never enough

Start with the structural fact that makes this whole topic confusing: IVF is not one event, it is a chain of events, and a success rate can be measured at any link in that chain. A treatment begins when a patient starts hormonal stimulation. It continues through an egg retrieval, then fertilization in the lab, then the growth of embryos over several days, then the transfer of an embryo into the uterus, then implantation, then a pregnancy test, then (if all goes well) an ongoing pregnancy, and finally a delivery many months later. A clinic can count "success" at the pregnancy test, at the transfer, at the retrieval, or across an entire patient's journey through multiple attempts. Each choice produces a different number from exactly the same set of patients.

Because the chain has so many links, a headline rate with no context is doing something subtle: it is asking you to trust that the clinic picked the link most relevant to your decision, rather than the link that makes the clinic look best. Those are rarely the same link. The number that flatters a clinic is the one measured latest in the chain (a pregnancy test is easier to hit than a delivery) and narrowest in the population (young patients with good prognoses succeed more often than the average person walking in the door). A number that helps you decide is the one measured at the outcome you care about, live birth, for a patient whose age and situation resemble yours.

It helps to picture IVF as a funnel, and each success rate as a measurement taken at one level of it. At the wide top are all the cycles that start. At each level below, some attempts drop out: cycles get cancelled, retrievals yield no usable embryo, transfers fail to implant, pregnancies miscarry. A rate measured near the top of the funnel divides by a big, inclusive group and tends to look modest. A rate measured near the bottom divides by the survivors, the attempts that already cleared every earlier hurdle, and tends to look impressive. Neither is lying, but they are answering different questions. A rate quoted without telling you which level of the funnel it was taken at is asking you to guess, and it is a safe bet that the level chosen was not the unflattering one. Once you train yourself to ask which level, the funnel stops being a marketing device and becomes a map.

This is why professional bodies publish success rates as large tables broken out by age and by stage, not as one figure. The federal data system exists precisely so that no clinic can reduce its performance to a single slogan. When you see a lone percentage in a brochure, the correct first reaction is not "is that good?" but "the top and bottom of what?" The rest of this guide is a tour of the answers, because once you can name the denominator and the numerator, the marketing loses most of its power.

There is a second reason one number fails you, and it is emotional rather than statistical. Fertility treatment is expensive, physically demanding, and time-sensitive, and the people reading these numbers are often making the most important decision of their lives under real time pressure as their own fertility declines. That combination makes a big, simple percentage extraordinarily persuasive, which is exactly why a big, simple percentage deserves the most scrutiny. The more a number is designed to reassure, the more you should ask how it was constructed. You can browse how these figures look side by side, cited to their source, on our clinics and research pages, but the skill you need first is the reading skill, so let us build it.

2. The four denominators, and why they disagree

The denominator is the bottom of the fraction: the group of attempts or people you are dividing your successes by. In US reporting there are four denominators in common use, and they produce numbers that can differ by more than double for the identical clinic and the identical patients. Understanding these four is the single most valuable thing in this guide, so it is worth slowing down. The four are: per cycle started, per embryo transfer, per intended egg retrieval, and per patient (cumulative). They are ordered here from the one that usually produces the lowest number to the one that usually produces the highest.

Per cycle started counts every cycle that begins stimulation, including cycles that get cancelled before retrieval and retrievals that never yield a usable embryo to transfer. It is the most conservative and, arguably, the most honest reflection of what happens when you sign up, because it does not quietly drop the attempts that failed early. Per embryo transfer only counts cycles that actually reached the point of putting an embryo back, which excludes everyone whose cycle was cancelled or who had no embryo to transfer. That exclusion is why per-transfer numbers look higher: the hardest cases have already been removed from the bottom of the fraction before you divide.

Per intended egg retrieval is the metric the Society for Assisted Reproductive Technology treats as its primary success measure, and it is the fairest of the cycle-based numbers. It counts everyone who set out to have a retrieval, and it credits them with a live birth if any embryo from that retrieval, transferred fresh or frozen later, results in a baby. This is powerful because modern IVF often produces several embryos from one retrieval, and a patient may transfer them one at a time over months. Measuring per intended retrieval captures the full value of that one egg collection instead of splitting it into separate transfer events. Nationally in 2022, live births per intended egg retrieval with a patient's own eggs were about 43.1% under age 35, 31% at ages 35 to 37, and 19% at ages 38 to 40 - Contemporary OB/GYN.

Per patient, cumulative is the highest number of all, because it follows a person across every cycle and every transfer they undergo until they either have a baby or stop. Because it adds up many chances, it produces the most optimistic figure. Reported per new patient, national 2022 outcomes were roughly 69% under 35 and about 24% at ages 41 to 42 - Illume Fertility, citing SART. Notice the gap: for the same young patients, the honest range runs from about 40% (per transfer or per retrieval) to about 69% (cumulative). A clinic that advertises "up to 69%" and a clinic that reports "43%" may be describing an identical practice.

There is a catch inside the cumulative number that patients should hold onto: it is only achievable if you actually complete all the cycles it assumes. A cumulative rate that reaches, say, 65% might be built on the average patient undergoing two or three retrievals, which means it quietly assumes the time, the money, and the physical tolerance for multiple rounds. For a patient who can afford one cycle, the cumulative figure overstates the realistic chance considerably, because it credits attempts that person may never make. This does not make the cumulative number dishonest, but it does make it conditional, and the condition is often unstated. When you see a cumulative or per-patient rate, the follow-up question is "across how many cycles," because a 65% chance across three rounds is a very different promise from a 65% chance on the first try, and only one of them may be available to you.

It is worth dwelling on why the per-intended-retrieval measure is considered the fairest, because the reasoning reveals how modern IVF actually works. A single stimulation and retrieval commonly yields more than one viable embryo, and clinics increasingly freeze all of them and transfer them one at a time to avoid the risks of putting several back at once. If you counted each of those frozen transfers as its own separate attempt, a retrieval that produced three embryos and eventually one baby could be reported as one success out of three transfers, or the successful transfer could be reported in isolation as a strong standalone result. Per intended retrieval sidesteps both distortions by asking a cleaner question: of everyone who set out to collect eggs, how many ended up with a baby from that collection, counting every fresh and frozen transfer it produced. That is why SART treats live birth per intended retrieval, not per transfer, as its primary outcome.

The gap between per cycle started and per intended retrieval is usually small, but it is not zero, and it lives in the cancelled cycles. A cycle can be called off before retrieval when a patient responds poorly to stimulation, producing too few follicles to justify the procedure and its anesthesia, and whether those cancelled cycles sit in the denominator depends on the exact metric being quoted. A clinic that reports only outcomes from completed retrievals has, once again, removed a set of harder cases from the bottom of the fraction. None of this is hidden if you read the definitions, which is precisely why reading the definitions is the habit that protects you.

To make the mechanics concrete, here is a deliberately simplified worked example. It uses round hypothetical numbers to show how one reality yields four rates. It is an illustration of the arithmetic, not real clinic data.

Every rate in that table is truthful and every one describes the same 40 babies. The only thing that changed was the group on the bottom. This is why the practical rule is simple and strict: never accept a success rate until you know its denominator. When you compare two clinics or two brochures, the first job is to confirm they are quoting the same denominator, because a per-transfer number next to a per-cycle number is not a comparison at all, it is a category error dressed up as one. Our approach on IVFcost.co is to take clinic outcomes from the CDC's own reporting and label the denominator on every figure, so you are never quietly handed the flattering one.

3. The numerator problem: a positive test is not a baby

The denominator gets most of the attention, but the numerator (the top of the fraction, what counts as a "success") hides its own trap. There is a meaningful clinical distance between a positive pregnancy test, a clinical pregnancy confirmed by ultrasound, an ongoing pregnancy, and a live birth. Each of those is a legitimate thing to measure, and each attrites into the next: not every positive test becomes a clinical pregnancy, and not every clinical pregnancy becomes a delivery. A clinic that reports "pregnancy rate" is measuring a numerator that sits earlier and higher in the chain than a clinic reporting "live birth rate," and the gap between them is real pregnancies that ended in miscarriage.

This is why the serious data systems insist on live birth, or more precisely live-birth delivery, as the outcome that matters. SART states plainly that it emphasizes the delivery of a child rather than a positive pregnancy test as the main outcome of interest, because a positive test is a hope and a delivery is a baby. The distinction is not pedantic for older patients in particular, where miscarriage rates are higher, so the distance between "got pregnant" and "had a baby" widens with age. A pregnancy rate quoted without an age band and without a live-birth figure beside it is one of the softest numbers in the entire field, and it should be read with corresponding caution.

The vocabulary here is worth learning because clinics use it precisely even when marketing uses it loosely. A biochemical pregnancy is a positive blood or urine test showing the hormone hCG and nothing more: it can, and often does, resolve into nothing. A clinical pregnancy is one confirmed on ultrasound, typically with a gestational sac and later a fetal heartbeat, which is a stronger signal but still not a guarantee. An ongoing pregnancy has progressed past the highest-risk early weeks. A live birth is a delivered baby. Each of these is a real, defined milestone, and each attrites into the next, so a rate built on the earliest milestone will always be the largest. When a page says "pregnancy" without specifying which kind, it is safe to assume the most generous interpretation is the one being shown.

Miscarriage is the specific reason the pregnancy-to-birth gap is not constant but grows with age. Because the likelihood that an embryo carries the wrong number of chromosomes rises as eggs age, older patients both conceive less often and lose more of the pregnancies they do achieve. That double effect means a 42-year-old reading a bare "pregnancy rate" is being shown a number whose distance from an actual live birth is wider than it would be for a 32-year-old reading the same label. An age-blind pregnancy rate is therefore doubly misleading for older patients: it inflates the top of the chain and then hides the age at which the inflation is largest. The defense is the same one this guide keeps returning to, which is to refuse the earlier-stage number and hold out for live birth in your own age band.

Be alert to the exact words. "Pregnancy rate," "positive beta," "implantation rate," and "clinical pregnancy rate" all describe links earlier than birth. "Live birth rate" and "live-birth delivery rate" describe the outcome you are actually paying for. When a marketing page leads with a pregnancy rate and buries or omits the live birth rate, that ordering is a choice, and it is usually the choice that produces the larger headline. The fix is to hold out for the live birth figure and to treat any earlier-stage number as context, not conclusion.

There is one more numerator subtlety worth naming, because it cuts the other way and is genuinely in patients' interest. Twins and higher-order multiples inflate the count of babies but carry serious health risks for both the pregnant person and the infants, so the field has spent a decade deliberately moving toward transferring a single embryo at a time. A clinic with a slightly lower pregnancy rate but a much lower multiple rate may be practicing better medicine, not worse. When you read a numerator, then, ask two questions rather than one: is this counting births or something earlier, and is this counting healthy singleton births or padding the total with risky multiples. The best outcome is a single, healthy, full-term baby, and a rate that quietly rewards multiples is optimizing for the wrong thing.

4. Age is the filter that changes everything

If you remember only one thing from this guide, remember that female age is the dominant variable in IVF outcomes, and that any success rate not broken out by age is hiding its most important dimension. The reason is biological rather than procedural. As a person ages, the number of eggs falls and, more decisively, the proportion of eggs with the correct number of chromosomes falls. Chromosomally abnormal embryos are far less likely to implant and far more likely to miscarry. No laboratory technique reverses this, which is why the age gradient in the national data is so steep and so consistent year after year.

The mechanism deserves a sentence of its own, because it explains why no amount of clinic skill overrides it. Eggs are held in a kind of biological suspension from before birth, and the cellular machinery that divides their chromosomes correctly degrades over the decades. By a woman's forties, a large majority of her eggs carry an abnormal chromosome count, a state called aneuploidy, and an aneuploid embryo usually either fails to implant or is lost to early miscarriage. A laboratory can select the best embryo available, culture it superbly, and transfer it perfectly, but it cannot manufacture a chromosomally normal embryo from eggs that no longer reliably produce one. That is the hard floor beneath every older patient's odds, and it is why the single most powerful lever in IVF outcomes is one no clinic controls.

Age being dominant does not mean age is the only variable, and a good reader holds both ideas at once. Sperm quality, the specific cause of a couple's infertility, uterine conditions such as fibroids or scarring, embryo culture quality in the lab, and simple chance all move an individual outcome. This is why two people of the same age can have very different results, and why your own number is never fully knowable in advance from a table. The national age bands are the strongest single predictor and the right anchor for expectations, but they are a starting point for a conversation with a clinician who can weigh your other factors, not a verdict. Reading rates well means taking age seriously without treating it as destiny.

The numbers make the gradient vivid. Using the fairest cycle-based denominator, per intended egg retrieval with a patient's own eggs, the national 2022 live birth rate falls from about 43% under 35 to 19% at 38 to 40, and to roughly 3.2% for patients over 42 - Contemporary OB/GYN. The American Society for Reproductive Medicine puts the clinical bottom line even more directly: by age 43, the chance of becoming pregnant with a woman's own eggs through IVF is less than 5%, and by the mid-forties donor eggs become, for most people, the only realistic path - ASRM ReproductiveFacts. That is not a marketing statement, it is the professional society describing the biology.

At the oldest ages the story is not only that rates are low but that additional cycles stop helping. A large analysis of SART data covering women over 42 found that the cumulative live birth rate, the running total across every cycle a patient attempts, plateaus after very few cycles: women at 43 and 44 saw little gain beyond the fifth cycle, women 45 to 46 plateaued by the third, and women 47 and older gained essentially nothing after the first - Cumulative live birth rates over age 42 (PMC). The maximum cumulative rate itself dropped sharply with each year of age, as the chart below shows.

It also helps to separate two things that patients, and even some marketing, conflate: how many eggs you have versus how good they are. Ovarian reserve tests such as anti-Mullerian hormone levels and antral follicle counts predict roughly how many eggs a stimulation cycle will yield, which matters for planning, but they say little about the chromosomal quality of those eggs, which is what age governs. A younger patient with a low egg count and an older patient with a high one are not in the same position, because the younger patient's fewer eggs are more likely to be normal. This is why age bands, not reserve numbers, anchor the national success tables, and why a clinic that emphasizes a reassuring reserve result without discussing age may be answering the easier question rather than the decisive one. The large SART analysis underlines the point at scale: across roughly 24,650 women over 42 and more than 58,000 cycles, the path to a live birth was short and steep, not long and hopeful.

The practical consequence for reading rates is direct. When you look at any clinic's figure, find the row for your own age band and ignore the rest, because the all-ages average is a blend that describes no real person. The standard bands used in US reporting are under 35, 35 to 37, 38 to 40, 41 to 42, and over 42, and they exist because outcomes shift meaningfully across each of them. A clinic that advertises a single headline rate without letting you see your band is not necessarily hiding something, but it is definitely not helping you, and the fix is to go to the underlying CDC or SART report, which always breaks the numbers out by age. You can also start from an age-aware view on our research and success-rate pages, where the age dimension is never collapsed into a single figure.

The age gradient cuts in a hopeful direction too, and it is worth stating plainly so the message is not only cautionary. For patients under 35, the national numbers are genuinely encouraging, and even the fair per-retrieval denominator sits above 40%, with cumulative chances across a few cycles higher still. Reading rates well is not about lowering everyone's expectations, it is about matching expectations to reality, and for younger patients the reality is often better than a conservative single-cycle number first suggests. The same discipline that protects an older patient from an inflated headline protects a younger one from unnecessary despair, because it points both of them to the specific, honest figure for their situation rather than to a fear or a hope.

5. Patient mix: how who a clinic treats moves its number

Here is the most counterintuitive idea in the whole subject: a clinic can raise its published success rate without improving its medicine at all, simply by changing which patients it treats. This is called patient mix or case mix, and it is the reason the regulators warn so strongly against comparing clinics by their headline numbers. The mechanism is pure arithmetic. If a clinic tends to accept younger patients, patients with good ovarian reserve, and patients without a history of failed cycles, its average outcome will be high because the underlying prognosis of its patient pool is high, independent of anything the clinic does in the lab.

Consider two clinics that are genuinely equal in skill. Clinic A accepts everyone, including women over 42, patients with very low egg counts, and patients who have already failed several cycles elsewhere. Clinic B quietly steers those same patients away, perhaps by requiring minimum egg-reserve thresholds or discouraging poor-prognosis patients at the consultation stage. Clinic B will post better numbers every single year, not because its embryologists are better, but because it removed the hardest cases from its own denominator before treatment even started. A patient reading only the headline would wrongly conclude that Clinic B is the stronger choice, when in fact Clinic B might be the one that would turn them away.

You can even see the fingerprint of patient mix in a few secondary numbers, if a report exposes them. A clinic that treats mostly favorable cases will tend to show a low cancellation rate, a low share of patients over 40, a high average number of eggs retrieved, and few cycles coded as poor-responder or diminished-reserve. A clinic that takes hard cases will show the opposite pattern, and its headline live birth rate will be lower even if its per-case skill is identical or better. None of these secondary figures is a scandal on its own, but read together they sketch what kind of population produced the headline, which is exactly the context the headline itself omits. This is the difference between reading a number and reading around it.

Geography and insurance quietly shape patient mix too, in ways that can make the most accessible clinics look statistically worse. In places where IVF is covered by an insurance mandate, more people can afford treatment, including older patients and those with difficult prognoses who would otherwise be priced out, which broadens the population a clinic treats and pulls its average outcome down. In places where patients pay entirely out of pocket, cost itself filters the population toward those who are younger, wealthier, or more determined, which can lift a clinic's average without any difference in care. So a lower headline rate can sometimes be a sign of broader access rather than weaker medicine, and this is one more reason the raw number cannot be read as a quality score. The context that produced it is doing a great deal of the work.

This is not hypothetical, and the professional bodies say so out loud. SART's own reports carry the warning that the data should not be used to compare clinics because clinics differ in patient selection and treatment approaches that can artificially inflate or lower rates relative to one another - SART National Summary. The unavoidable implication is that a very high advertised rate can be a signal about who a clinic accepts rather than how well it performs, and that the most rigorous clinics, the ones willing to take on difficult cases, may show lower numbers precisely because they are doing the harder and more valuable work.

So how do you read around patient mix? You cannot fully correct for it from the outside, but you can defend yourself. Anchor on your own age band and, where the report allows, on sub-populations that resemble you, such as patients with a diagnosis like yours or patients on their first cycle. Compare a clinic to the national average for your band rather than to another clinic's flattering headline. And treat an unusually high rate with curiosity rather than awe: ask whether the clinic publishes the share of patients it declines, what its cancellation rate looks like, and how it handles poor-prognosis cases. A clinic proud of its outcomes should be comfortable answering. If the number is high because the population is easy, that is worth knowing before you commit your money, your body, and your time.

6. The comparison trap: why the regulators say do not compare clinics

It surprises most people to learn that the two organizations responsible for publishing US fertility outcomes both explicitly tell you not to use their data to rank clinics against each other. This is not modesty and it is not a legal disclaimer bolted on by lawyers. It is a direct consequence of the patient-mix problem in the previous section, combined with the reality that the reported numbers are not risk-adjusted in a way that makes a clean clinic-to-clinic comparison valid. Understanding why the warning exists tells you how to use the data correctly, which is a different and more useful thing than not using it at all.

The federal reporting system was created by the Fertility Clinic Success Rate and Certification Act of 1992, a law that requires every ART program in the country to report its pregnancy success rates to the Centers for Disease Control and Prevention in a standardized format, and requires the CDC to publish them - Congress.gov. The intent was consumer protection: before the law, clinics advertised wildly inconsistent and sometimes fabricated success claims, and patients had no neutral way to check them. The law fixed the fabrication problem by forcing a common definition and a public dataset. What it could not fix is that clinics still see different patients, so identical reporting can still reflect different underlying difficulty.

It is worth remembering what the world looked like before that law, because it explains the law's shape. In the 1980s and early 1990s, IVF was new, its advertising was unregulated, and it was lucrative, and some clinics promoted success figures that were inconsistent, non-comparable, or simply invented, with no shared definition of "success" and no neutral place to check a claim. Congress responded by mandating standardized reporting and public publication rather than by capping prices or dictating practice. Tellingly, the law made reporting mandatory but attached no financial penalty for a clinic that declines to participate, relying instead on the visibility of being listed as a non-reporting program. That design choice tells you how to weight the data: it is a near-complete public record because the overwhelming majority of clinics choose to report and be counted, not because a regulator forces every number into existence.

That is exactly why the CDC devotes a page to how to interpret its numbers and cautions against using them to compare programs, and why SART repeats the warning on every report - CDC, How to Interpret ART Success Rates. The data is excellent for two purposes and poor for a third. It is excellent for understanding realistic national expectations for someone your age, and excellent for sanity-checking a specific clinic's claim against the public record to confirm the clinic is not inventing numbers. It is poor for declaring Clinic B better than Clinic A on the strength of a few percentage points, because those points can be entirely explained by who each clinic treats. That said, the clinic-to-clinic spread in the national data is far too large to be all noise: our analysis of the CDC data on 457 clinics found that, for under-35 patients, the middle half of clinics span roughly 36% to 55% live births per retrieval, so once you have controlled for age and read each rate next to its cycle volume, where you go still matters.

The right mental model is to use the data as a floor and a filter rather than a leaderboard. Use it to rule out claims that do not match the public record, to set your own expectations honestly by age, and to generate specific questions for a consultation. Do not use it to crown a winner. When a comparison site, a clinic, or an advertisement ranks clinics purely by headline success rate and presents that ranking as authoritative, it is doing precisely the thing the data's own custodians tell you not to do. The honest version of comparison holds age, denominator, and outcome constant, discloses what it cannot adjust for, and never lets a raw rate override those cautions. That is the standard we try to hold on IVFcost.co, and it is the standard you should demand of any source.

7. Donor eggs quietly swap the denominator

One of the easiest ways for a headline number to mislead is to blend cycles using a patient's own eggs with cycles using donor eggs, because donor-egg outcomes are dramatically higher and largely disconnected from the recipient's age. Donor eggs typically come from young, screened donors, so they carry the favorable chromosomal profile of youth regardless of how old the recipient is. Nationally in 2022, transfers using donor eggs or donor embryos produced live-birth rates in the range of roughly 42% to 52% per transfer across all recipient ages - CDC ART Success Rates. Those numbers sit far above what a 44-year-old could expect from her own eggs.

The reason this matters for reading rates is that a clinic serving many older patients has a strong incentive, conscious or not, to let donor-egg successes lift its overall average. If a 45-year-old sees a clinic advertising a rate that looks encouragingly high for her age, she needs to know whether that figure includes donor-egg cycles, because if it does, it is describing a different treatment than the one she may be planning. The high success of egg donation is itself clinically informative: it confirms that egg quality, not uterine capacity, is the primary age-related barrier for most women, which is precisely why donor eggs work so well later in life - ASRM ReproductiveFacts.

Good reporting keeps these two worlds separate, and so should you. When you read a clinic's numbers, confirm you are looking at the own-egg (sometimes called autologous or nondonor) figures if you plan to use your own eggs, and at donor figures only if that is your path. The federal reports separate them deliberately, presenting nondonor cycles stratified by the patient's own age and donor cycles as their own category, because merging them would destroy the meaning of both. A rate that does not tell you whether donor eggs are included is a rate you cannot use, and the correct response is to find the underlying report where the two are broken apart.

There is a further wrinkle inside donor cycles, which is that donor eggs themselves come fresh or frozen, and the embryo transferred can be fresh or thawed, and each combination reports slightly differently. That is why the national donor figure is a range rather than a single number: the roughly 42% to 52% band reflects those different circumstances, not uncertainty about the data. For a patient weighing donor eggs, the useful move is to read the specific sub-category that matches the plan being proposed, rather than the blended donor average, in exactly the same spirit as reading your own age band rather than the all-ages number. The principle does not change when the treatment does: match the figure to the choice in front of you.

One honest caveat belongs alongside the encouraging donor-egg numbers, because a success rate is not the whole picture of a decision. While donor eggs restore the egg-quality odds of youth, they do not make the pregnancy itself risk-free for an older recipient: carrying a pregnancy at an advanced age still raises the chance of complications such as high blood pressure and gestational diabetes, which is a medical conversation separate from the live-birth percentage. A number can tell you the probability of a birth without telling you the risks along the way, and a responsible reading of donor-egg rates keeps those two questions distinct. The point is not to discourage anyone, it is to notice that even a well-labeled, honest success rate answers only the question it was built to answer, and the rest belongs with your clinician.

This also reframes a decision many patients face in their forties as a data-reading problem rather than only an emotional one. The choice between continuing with your own eggs at a low per-cycle chance and moving to donor eggs at a much higher chance is, in part, a choice between two very different denominators and numerators, and seeing the real numbers for each, side by side and correctly labeled, is what makes the decision an informed one. Costs diverge sharply between the two paths as well, which is why it helps to read success and price together rather than in isolation. Our costs and providers pages exist to put those two dimensions next to each other, but the reading principle stands on its own: never let a donor-inflated average masquerade as an own-egg expectation.

8. Cycle counts, cancellations, and batching

Beyond age and patient mix, there are structural choices in how a clinic runs and reports cycles that move its numbers, and knowing them helps you read a rate for what it is. The first is cancellation. A cycle that is started and then cancelled before egg retrieval, often because the patient is responding poorly to stimulation, may or may not appear in the denominator depending on which metric is quoted. A clinic that reports only per-transfer rates has effectively removed every cancelled cycle and every retrieval that produced no transferable embryo from the bottom of its fraction, which mechanically raises the number without any change in outcomes for the patients who were dropped.

It is useful to know that the number of eggs retrieved and embryos created is itself a kind of early success signal, and one that clinics sometimes emphasize because it arrives long before a live birth is known. More eggs and more embryos generally mean more chances, and for younger patients a higher egg yield does correlate with a better cumulative outcome. But egg count is an input, not an outcome, and it is easy to present a strong average egg number as though it were a strong success number. A patient can produce many eggs and still not have a baby, especially at older ages where quantity does not fix quality. So when a clinic leads with how many eggs it typically retrieves, treat that as context about the process, not as evidence about the result, and keep your eyes on the live-birth figure for your band.

The second structural choice is how frozen embryos are counted. Modern IVF frequently freezes all embryos from a retrieval and transfers them later, sometimes months later, in separate frozen embryo transfer cycles. Depending on the metric, those frozen transfers can be credited back to the original retrieval (the per-intended-retrieval approach, which is fair) or counted as their own fresh set of attempts (which can make per-transfer numbers look better because frozen single-embryo transfers in good candidates succeed at high rates). This is why the per-intended-retrieval denominator is the one professional reporting favors: it prevents a single egg collection from being sliced into several separately-counted transfer events, each of which looks like a strong standalone success.

The freeze-all strategy has a legitimate clinical rationale that is worth understanding, because it shows how a sound medical decision and a reporting effect can travel together. Freezing every embryo and transferring in a later cycle lets the uterine lining recover from the intensity of stimulation and be prepared on a cleaner schedule, which many clinics believe improves implantation. That is a reasonable choice made for the patient's benefit. But the same choice also shifts where successes land in the reporting, moving them out of the fresh-transfer column and into frozen cycles that, measured on their own, look excellent. The lesson is not that freeze-all is a trick, it is that a clinic's clinical style and its reported numbers are entangled, so you should read the per-retrieval figure, which is neutral to that style, rather than a fresh-only or frozen-only slice that the style happens to flatter.

A third practice, less about deception and more about interpretation, is batching and case selection at the margins. Clinics differ in how aggressively they stimulate, when they choose to cancel, whether they encourage genetic testing of embryos before transfer, and how they counsel poor responders. Each of these reasonable clinical decisions nudges the reported rate. None of that is dishonest, but all of it means two clinics reporting the same denominator can still differ because their clinical pathways differ, which is one more reason to treat small gaps between clinics as noise rather than signal.

Preimplantation genetic testing for aneuploidy, often shortened to PGT-A, is the clearest example of a practice that reshapes a rate in two directions at once. By screening embryos and transferring only those with a normal chromosome count, a clinic can raise its live birth rate per transfer, because the transfers that do happen start from a stronger embryo. But the same screening reduces the number of transfers that occur at all, and for some patients, especially those with few embryos, it can lower the overall chance of a live birth per retrieval by setting aside embryos that might have worked. Whether PGT-A helps a given patient is genuinely debated among specialists, but for the narrow purpose of reading rates the lesson is simple: a strong per-transfer number at a clinic that tests heavily is partly a story about selection, not only about the lab, and it should be read next to the per-retrieval number to see the fuller picture.

The reading strategy that survives all of this is to prefer the denominators that are hardest to game and to distrust the ones that are easiest. Per cycle started and per intended retrieval are difficult to inflate because they hold on to the failures and the early exits. Per transfer is the easiest to inflate because it silently drops them. So when a clinic leads with a per-transfer number, the useful move is to ask for the per-intended-retrieval or per-cycle figure for your age band and watch how far it drops. A large drop tells you the headline was buoyed by everyone who never made it to transfer. A small drop tells you the headline was reasonably representative. Either way, you learn something the brochure did not want to volunteer.

9. Reading a SART or CDC report line by line

Once you know the concepts, the mechanics of reading an actual report are straightforward, and doing it yourself is the best protection against a cherry-picked quote. Both the CDC and SART publish clinic-level and national data online for free. The CDC's system, built under the 1992 law and run through the National ART Surveillance System, captures an estimated 97% to 98% of all ART cycles performed in the country, which makes it close to a full census rather than a sample - CDC NASS Technical Notes. SART, the professional society, publishes an overlapping dataset drawn from its member clinics, which perform the large majority of US cycles - SART National Summary.

A word on the two datasets, because patients often wonder which to trust. The CDC system is the legal one, built to capture essentially every clinic in the country, which is what lets it function as a near-census. SART is the professional society, and its public reports draw from its member clinics, which perform most but not all US cycles and which agree to a shared reporting standard and validation. The two overlap heavily and tell the same story, so you do not need to choose: use whichever presents your clinic and your age band most clearly, and cross-check a surprising figure against the other. What you should not do is accept a clinic's own website figure in place of either, because the website number has not necessarily passed through the validation both public systems apply.

When you open a clinic report, read it in a fixed order so the presentation cannot steer you. First, find the reporting year, because the data is always a few years old and you want to know how old. Second, find the own-egg versus donor-egg split and stay in the column that matches your plan. Third, find your age band and ignore every other band. Fourth, identify the denominator on the specific number you are reading, whether it is per cycle, per transfer, or per intended retrieval. Fifth, confirm the outcome is live birth and not a pregnancy stage. If you do those five things, in that order, you will be reading the same figure a specialist would read, and no headline can substitute a friendlier number for the one that applies to you.

Here is what that looks like in practice for a hypothetical 37-year-old planning to use her own eggs. She opens the clinic's public report, confirms it covers a recent year, and moves to the nondonor section, ignoring the donor column entirely. She finds the 35-to-37 row and reads only that line. She sees a per-transfer number and, knowing it is the easiest denominator to inflate, looks for the per-intended-retrieval figure in the same row, which is lower and more honest. She confirms the outcome is live birth, not clinical pregnancy. Finally, she compares that figure to the national 35-to-37 benchmark rather than to a competitor's billboard. In five minutes she has replaced a marketing percentage with a number that actually describes her, and she walks into her consultation ready to ask why the clinic's figure sits where it does relative to the national one.

There is one more habit that separates a careful reader from a casual one: always compare a clinic to the national benchmark for your exact band, not to another clinic. The national number is the honest yardstick because it is built from nearly every cycle in the country and is not shaped by any single clinic's patient selection. If a clinic sits well above the national figure for your age, that is interesting but not conclusive, and it should prompt questions about patient mix rather than immediate confidence. If it sits well below, that is also worth a conversation. The benchmark turns a lone percentage into a comparison that actually means something.

Finally, treat the official report as the source of truth and every brochure, ad, or third-party summary as a claim to be checked against it. If a clinic's marketing figure cannot be found in, or reconciled with, its CDC or SART report, that discrepancy is itself information. The public dataset was created specifically so that patients could perform this check, and performing it is neither rude nor unusual. It is the entire point of the law. When we surface clinic outcomes on IVFcost.co, we tie each figure back to that same public reporting, so the check is already done and visible, but the skill of doing it yourself is one no marketing page can take away from you.

10. The reporting lag: what "2022 data" actually means

A detail that trips up even careful readers is that fertility success data is always old, and it is old for a good and unavoidable reason. To count a live birth, you have to wait for the birth. A cycle that starts in January must be followed through a retrieval, through possible frozen transfers months later, through a full pregnancy, and through delivery, which alone pushes the earliest possible complete data almost two years past the start of the cycle. Then the CDC validates the submissions through audits and site visits before publishing, which adds more time. The result is that the most recent complete, final national dataset typically describes cycles from two to three years earlier.

This lag has a practical implication for how you read a "latest" figure. As of the mid-2020s, the most recent complete national summary describes 2022 cycles, and clinic-level validated rates are published on a similar delay - CDC ART Surveillance. That does not make the data useless, because outcomes for a given age band move slowly from year to year and the biology of age does not change. But it does mean two things: a clinic that opened recently may have little or no published track record yet, and a clinic quoting its own very recent internal numbers is quoting figures that have not been through the same validation as the public dataset. Recent in-house numbers are not necessarily wrong, but they are not audited, and you should weight them accordingly.

This raises a fair question about new or fast-growing clinics, which may have opened after the most recent published year and therefore have little or no validated public record. The absence of data is not evidence of poor quality, but it does mean you are relying more on the credentials of the individual physicians and embryologists, the lab's accreditation, and the clinic's transparency than on a track record you can check. In that situation, the honest move is to say so to yourself: you are making a decision with less public information than usual, and you should compensate by asking more direct questions and by weighting the national benchmarks, which still describe what is achievable for your age, more heavily than any in-house claim the clinic makes about itself.

The lag also interacts with the reporting rules, which evolve. The federal requirements for what clinics must submit have been revised over the years, most recently through a formal rulemaking that updated the data fields and reporting obligations - Federal Register, 2022. When definitions change, comparing a very old report to a very new one can introduce small inconsistencies that have nothing to do with clinic performance. For everyday reading this is a minor effect, but it is one more reason to stay within a single recent reporting year and a single denominator when you compare anything.

Understanding the lag protects you from two opposite errors. The first is dismissing the official data as "out of date" and trusting a clinic's fresher-sounding marketing instead, which trades a validated near-census for an unaudited in-house claim. The second is treating a two-year-old number as a promise about today, when a clinic's staff, lab, and patient mix may all have shifted. The balanced reading is to use the validated public data as your anchor for realistic expectations, treat any clinic's newer internal figures as supplementary and unverified, and remember that the number describing your age band has been remarkably stable across years precisely because it is driven by biology that does not care what year it is.

11. The marketing playbook: how a number is made to look bigger

Having built the concepts, it is worth naming the specific moves that turn an ordinary clinic into an extraordinary-looking one on paper, because recognizing a move is what disarms it. None of these requires lying. Each one is a truthful number selected or framed to maximize impression, and each maps directly onto a concept from the earlier sections. Seeing them listed together is the closest thing to a decoder ring for fertility advertising.

The most common moves are these, and each has a one-line defense:

• Quoting the friendliest denominator. A per-transfer or cumulative "up to" figure instead of per cycle. Defense: ask for the per-intended-retrieval or per-cycle number for your age band.
• Dropping the age band. A single all-ages headline that blends young and old. Defense: refuse the average and read only your own band.
• Blending donor and own eggs. A high overall rate lifted by donor cycles. Defense: confirm you are in the own-egg column if that is your plan.
• Leading with pregnancy, not birth. A "pregnancy rate" that sits earlier in the chain than a live birth. Defense: hold out for the live birth figure.
• Comparing to a rival, not the benchmark. A ranking that ignores patient mix. Defense: compare each clinic to the national number for your band.

Each of those tactics is individually legal and often not even intended as manipulation, because the person writing the brochure genuinely believes their clinic is excellent and is simply presenting the flattering true number. That is what makes the pattern durable: it does not require bad actors, only the ordinary human tendency to lead with your best figure. But the cumulative effect on a patient who does not know the concepts is a systematically inflated sense of the odds, which can drive expensive decisions built on a number that never applied to them in the first place.

Picture how this lands on a real person. A 41-year-old sees a clinic advertise a "60% success rate" in large type. Unpacked, that figure turns out to be a cumulative, all-ages, all-egg-source, pregnancy-not-birth number, which is to say it combines the friendliest choice at every one of the levers this guide has described. Her actual realistic live-birth chance per retrieval with her own eggs, read from the national table for her band, is a small fraction of that. She is not the victim of a lie, because every component of the 60% is technically true for someone. She is the victim of compression: several honest choices stacked in the same flattering direction until the result no longer describes her at all. Decompressing that number is not cynicism, it is the difference between planning around 60% and planning around the truth.

The deeper point, reasoning from first principles, is that a success rate is an act of compression: it squeezes a complicated, multi-stage, age-dependent process into a single percentage, and every act of compression throws information away. The question is always which information got discarded and whether its absence favors the clinic. When you decompress the number by asking about denominator, age, egg source, and outcome, you are simply putting back the information that compression removed. A clinic that welcomes that decompression is showing you something reassuring about how it thinks. A clinic that resists it, or cannot answer, is showing you something too.

In fact, transparency itself is one of the few quality signals a patient can read from the outside. A clinic that publishes its rates by age and denominator, states clearly whether donor cycles are included, and can explain its cancellation and testing practices without defensiveness is telling you that it expects informed patients and is comfortable being measured honestly. That posture does not guarantee the best outcomes, but it correlates with the kind of rigor you want, and it is observable before you ever pay for a cycle. A clinic that offers only a single glossy percentage and grows uneasy when you ask what is underneath it has answered a different and equally useful question about how it prefers to be seen.

None of this means fertility clinics are adversaries or that high numbers are always suspect. Many clinics are genuinely excellent, report honestly, and would happily walk you through every denominator in the building. The purpose of learning the playbook is not cynicism, it is symmetry: the clinic knows exactly how its number was built, and you deserve to know too, so that the conversation happens between two informed parties rather than a marketer and a hopeful stranger. That symmetry is the whole reason a neutral, sourced reference exists, and it is the job we set for ourselves across IVFcost.co.

12. A practical checklist before you trust any rate

Everything above reduces to a short set of questions you can carry into any brochure, website, or consultation. Treat them as a sequence: each one strips away one layer of possible distortion, and by the end you are looking at a number that actually describes a person in your situation. The goal is not to catch anyone out. The goal is to convert a vague, persuasive percentage into a specific, honest expectation you can plan around.

Before you accept any IVF success rate, ask these five questions:

1. What is the denominator? Per cycle, per transfer, per intended retrieval, or cumulative per patient. If you cannot tell, the number is unusable until you can.
2. What age band is this? Find your own band and ignore the all-ages blend entirely.
3. Own eggs or donor eggs? Confirm the figure matches the treatment you actually plan to pursue.
4. Birth or pregnancy? Insist on live birth, and treat any earlier-stage number as context only.
5. Compared to what? Anchor the clinic against the national benchmark for your band, not against a rival's headline.

Work through those five and you will have done what the CDC and SART designed the public data to let you do, which is to hold any claim up against a validated national record and see whether it survives. The reporting system was built after decades of unregulated and sometimes fabricated advertising, and it exists so that no patient has to take a fertility number on faith - CDC National ART Surveillance System. The single most empowering fact in this entire field is that the real numbers are public, free, broken out by age, and only a few clicks away.

It is worth being honest that this checklist asks you to do work at the hardest possible moment, when you are hopeful, tired, and want a simple answer. That is real, and it is why having a source that has already done the unpacking matters. Whether you use our pages or read the CDC and SART reports directly, the aim is the same: to arrive at your consultation already knowing the realistic number for someone in your situation, so the conversation with your clinician is about your specific path forward rather than about decoding a percentage on a wall. An informed patient does not become a cynical one. They become a calmer one, because the numbers have stopped being a source of false hope or false despair and started being a tool.

The final principle is the one this whole guide has been building toward: a success rate is only as meaningful as the question it answers, and the question that matters to you is narrow and specific. Not "is this clinic good," but "what is the realistic live-birth chance for someone my age, using the eggs I plan to use, per honest attempt, at a clinic like this." When you make the question that precise, most misleading numbers simply fall away, because they were never answering it. And when you want to see those precise, age-aware, denominator-labeled figures gathered in one place and tied back to their public source, that is exactly what our clinics, costs, and research pages are for. Read the number, name its parts, and let the arithmetic, not the marketing, guide the most important decision you may ever make.

IVFcost.co is an informational data platform, not a medical provider, clinic, or laboratory. Nothing here is medical advice, a diagnosis, or a recommendation to start, stop, or change any treatment. Success rates depend heavily on individual circumstances and are not a promise of any outcome. National figures cited here reflect the most recent complete US reporting (2022 cycles) as published by the CDC and SART, and are current as of August 2026. Always confirm with a licensed clinician and the clinic's own current figures.