The Two IVF Success Rates (and Which One to Trust)

Every US fertility clinic reports two different IVF success rates for the same patients. Across 457 clinics, the two agree under 35 and split apart after it, by an average of 40% by age 40.

When a fertility clinic tells a 37-year-old that its IVF success rate is "around 40%," that number is almost always the live-birth rate per embryo transfer. It is a real figure, drawn from the same federal data the government publishes. It is also the more flattering of the two rates the clinic is required to report, because it quietly drops every cycle that never reached a transfer at all. The honest denominator, the one that answers the question a patient is actually asking (if I start a cycle, what are my odds?), is the live-birth rate per egg retrieval, and after age 35 it is materially lower.

We computed both rates for every CDC-reporting US fertility clinic in the most recent national ART success-rate data, the same dataset behind the clinic profiles on IVFcost.co. The gap between the two numbers is the single most useful thing a patient over 35 can understand before choosing a clinic, and almost no one is shown it. This is what the data says.

Key findings

• Under 35, the two rates are effectively identical. Across 457 clinics the average live-birth rate per retrieval and per transfer are both 40.8%, and at the typical clinic they sit within a point of each other (46.4% vs 47.1%). At this age, which number a clinic quotes barely matters.
• After 35, they split, and the advertised number wins. At ages 35-37 the average rate is 32.7% per transfer versus 27.2% per retrieval, and across clinics the per-transfer figure runs about 21% higher. By 38-40 the levels are 24.3% versus 17.2%, a per-transfer figure that averages roughly 40% higher clinic by clinic.
• The gap is the rule, not a few outliers. At 35-37, 275 of 457 clinics report a higher per-transfer than per-retrieval rate; at 38-40, 262 of 457 do.
• Over 40, the per-transfer rate mostly disappears. 316 of 457 clinics report zero or no live births per transfer for over-40 patients using their own eggs, because the samples are tiny. A single flattering number here is close to meaningless.
• This is not fraud, it is a denominator. Both rates are truthful. The difference is entirely in what each one counts, and clinics are free to lead with the one that reads best.

1. Two success rates, one procedure

IVF is not a single event, it is a chain: ovarian stimulation, an egg retrieval, fertilization in the lab, and finally one or more embryo transfers, sometimes months apart using frozen embryos. Because the chain can break at several points, there is no single natural place to measure "success." A cycle can be cancelled before retrieval, produce eggs but no viable embryo, or reach a freeze-all with a transfer that has not happened yet. Each of those is a real outcome for a real patient who spent real money, and each one is counted differently depending on which rate you look at.

The two rates the US requires clinics to report resolve this in opposite directions. Live birth per egg retrieval starts the clock at the retrieval and asks what share of those retrievals eventually produced a baby, so it includes the cycles that never reached a transfer. Live birth per transfer starts the clock later, at the moment an embryo is actually placed, so it only ever divides by cycles that got that far. The federal methodology behind both is spelled out in the CDC's technical notes, and the distinction is not a technicality: it decides whether a failed stimulation or an abandoned cycle drags the number down or vanishes from it entirely.

Which denominator is "right" depends on the decision in front of you:

• Deciding whether to start IVF at all: per retrieval (or per cycle started) is the honest measure, because you are standing before the whole chain, not at the transfer.
• Deciding how many embryos to transfer once you already have them: per transfer is the relevant number.
• Comparing your realistic odds across clinics: per retrieval, because it is far harder to flatter by patient selection.

This is why the Society for Assisted Reproductive Technology, the body clinics report through, emphasizes per-cycle outcomes over per-transfer ones: counting each cycle rather than each transfer, as one SART-focused clinical explainer puts it, "more accurately reflects the treatment burden and costs the patient has to potentially endure." A patient deciding whether to spend $20,000 or more on a cycle is asking a per-retrieval question. The advertising answers a per-transfer one. Our own guide on how to read IVF success rates walks through the same trap from the patient's side.

2. What the CDC data on 457 clinics shows

To see how much the choice of denominator actually changes the picture, we took both rates for all 457 CDC-reporting fertility clinics and averaged them within each of the four age bands the CDC publishes. The result is a clean, consistent pattern that holds across the whole national dataset rather than at any single clinic. Under 35, the two lines are indistinguishable. From 35 onward they separate, and the separation widens with every age band.

The chart below is the entire argument of this report in one image. The blue bars are the per-retrieval rate (the whole-chain number) and the second series is the per-transfer rate (the advertised number). Notice that they start on top of each other and then fan apart.

The under-35 result is worth pausing on, because it is the control case. When a patient is young enough that most retrievals produce a transferable embryo and most cycles reach transfer, the two denominators are nearly the same set of cycles, so the two rates converge: 40.8% either way on average, and 46.4% versus 47.1% at the median clinic. At this age, a clinic that quotes its per-transfer rate is not really inflating anything, because there is little to inflate. This matters because it rules out the innocent explanation for the older bands. The divergence after 35 is not an artifact of how we computed the numbers. It is what happens when more cycles start failing before they reach a transfer, and only one of the two rates notices.

The separation itself is stark. By 35-37 the average per-transfer rate is 5.5 points higher than per-retrieval, and by 38-40 it is 7.1 points higher, on a base that has itself fallen to the high teens. On a per-cycle basis this is the age band where the most consequential decisions get made, and it is exactly where the two published numbers disagree most in relative terms. For a fuller picture of how much clinic identity (not just age) moves these odds, our analysis of the CDC data on all 457 clinics found under-35 live-birth rates per retrieval ranging from near zero to 79.5%, a reminder that age and denominator are two of several axes a patient is rarely shown together.

3. How big the gap gets, and where

Averages can hide as much as they reveal, so the more important question is whether the gap is a broad, systematic feature of the data or the product of a handful of extreme clinics. It is systematic. When we compared the two rates within each individual clinic, the per-transfer rate came out higher at the majority of clinics in every age band above 35, not at a lucky few. The pattern is baked into the structure of the procedure, so it shows up almost everywhere the samples are large enough to see it.

Expressed as the relative amount by which the advertised per-transfer rate exceeds the honest per-retrieval rate, the inflation climbs steadily with age. It is essentially zero under 35, and then it grows into the tens of percent.

Behind those averages sit the clinic counts, which are the part that should change how a patient reads any single brochure. At 35-37, the per-transfer rate is the higher of the two numbers at 275 of 457 clinics. At 38-40, it is higher at 262 of 457. In other words, if you are in your late thirties and a clinic hands you one success percentage without saying which denominator it used, the base rate is that it picked the more flattering one, because at most clinics that number genuinely is more flattering. The tail is wide too: at the clinics where the gap is largest, the per-transfer figure runs more than 20 percentage points above the per-retrieval figure for the same patients.

None of this requires anyone to lie. A clinic can publish a completely accurate 38% per-transfer rate while its per-retrieval rate for the same age band is 27%, and both numbers pass an audit. The reader simply has to know which question each one answers. That is the whole reason IVFcost.co shows the per-retrieval rate as the headline figure on every clinic profile and labels the denominator explicitly, rather than letting the friendlier number stand in for the patient's real odds.

4. Why the advertised number is usually per transfer

Reasoning from first principles, the incentive is structural, not conspiratorial. A clinic competes for patients partly on its published outcomes, and of the two truthful numbers available, one is reliably higher after 35. Absent a rule forcing a specific denominator in patient-facing marketing, the higher number is the one that gets printed on the website, quoted on the phone, and framed in the lobby. You would expect this outcome from any set of clinics acting rationally, which is exactly why it shows up uniformly in the data rather than at a few bad actors.

There is also a legitimate clinical argument for the per-transfer rate, and honesty requires stating it. Per-transfer success partly reflects embryo-lab quality and transfer technique, the parts of the process most directly under the clinic's control, so a clinic can reasonably say it measures what the clinic itself does well. The problem is not that the number is meaningless, it is that it answers a narrower question than the patient is asking. A per-transfer rate tells you how good the clinic is at the last step. A per-retrieval rate tells you how likely the whole thing is to work. Those diverge most precisely when eggs and embryos get scarcer, which is to say, with age.

This is why the federal reporting exists at all. The 1992 Fertility Clinic Success Rate and Certification Act was passed because clinics were advertising success rates with no common definition, and patients had no way to compare like with like. The CDC's national ART reporting and the SART outcome tables standardized the measurement so that a per-retrieval rate at one clinic means the same thing as at another. The standardization does not, however, control which of the standardized numbers a clinic chooses to feature in its own marketing, and that is the gap this report is about.

5. The over-40 blind spot

The oldest age band deserves its own section, because here the per-transfer rate does not just flatter, it largely stops existing. Among the 457 clinics, 316 report zero or no live births per transfer for patients over 40 using their own eggs. That is not because those clinics never succeed with older patients. It is because the number of over-40 own-egg transfers at any single clinic in a reporting period is often tiny, and a denominator of a handful of transfers produces a rate that is either 0% or 100% and means almost nothing either way.

The CDC itself warns against exactly this misreading. Small case counts, differences in patient mix, the share of cycles using PGT-A embryo screening, and plain random chance can all swing a single clinic's rate dramatically, which is why the national data is more trustworthy in aggregate than clinic by clinic at the extremes. If only three embryos were transferred to over-40 patients at a clinic, the reported rate is a coin-flip artifact, not a track record. This is the same small-sample problem that makes the flashiest single-clinic numbers the least reliable, and it lands hardest on the patients who most need honest odds.

The practical takeaway for over-40 patients is to distrust any single clinic percentage and reason from the national picture instead. Averaged across all 457 clinics, the per-retrieval live-birth rate for over-40 own-egg IVF is under 6%, which is why so many clinics and patients turn to donor eggs or cumulative outcomes across multiple retrievals at this age. Our cost breakdown for IVF covers how those multi-cycle and donor paths change the real budget, since the per-cycle number and the per-baby number diverge fastest exactly here.

6. What to ask your clinic

The entire practical value of this analysis collapses into a few questions a patient can ask before signing anything, and they cost nothing. The goal is not to catch a clinic in a lie, since neither number is a lie, but to convert whatever figure you are quoted back into the honest denominator so you can compare clinics and set your own expectations on the same basis. A clinic acting in good faith will answer these plainly, and a clinic that dodges them has told you something too.

Ask for the numbers in the terms that actually describe your situation:

• "Is that rate per transfer or per retrieval?" If they do not volunteer it, this single question reframes the entire conversation.
• "What is your live-birth rate per egg retrieval for my age band?" This is the whole-chain number, and the one worth comparing across clinics.
• "What is the cumulative live-birth rate across all transfers from one retrieval?" This is the most patient-meaningful figure of all, and the IVF success estimator the CDC publishes is built around it.
• "How many cycles like mine did that number come from?" A rate built on a dozen cases is a hint, not a track record.

Once you have the per-retrieval figure, you can place any clinic against the national picture instead of against its own marketing. The clinic profiles on IVFcost.co already lead with the per-retrieval rate by age band and cite it to the CDC, precisely so that the comparison you are trying to make by hand is done for you. If you want to see how your own age and treatment plan translate into a realistic all-in budget across the cycles most patients actually need, our IVF cost calculator works from the same sourced figures.

Methodology

This analysis uses the most recent CDC Assisted Reproductive Technology (ART) clinic-level data for all 457 CDC-reporting US fertility clinics, restricted to cycles using a patient's own eggs. For each clinic we took the reported live-birth rate per egg retrieval and live-birth rate per embryo transfer in each of the CDC's four age bands (under 35, 35-37, 38-40, over 40). Averages are the mean across clinics within a band; "typical clinic" figures are medians. Within-clinic comparisons pair the two rates for the same clinic and age band. Rates for the over-40 band, and for any clinic with a small case count, are unstable by nature and are treated accordingly above. Every clinic figure on this site traces to its CDC source, as documented in our methodology.

This report reflects the CDC ART national data available as of August 2026. IVF success rates are published on an annual reporting cycle and are population statistics, not a prediction of any individual outcome. It is not medical advice. Verify the current figures for any specific clinic against its own CDC clinic report before making decisions.