Before the comments start: two things I'm not saying
One: You can absolutely do IVF if you want to. A choice without the full picture isn't a free choice. That's the point of informed consent and I don't think most people are getting it. Plus I believe no matter your choice for IVF, we should all aim to cover all bases to improve odds and avoid multiple rounds. & also be your healthiest self no matter what.
Two: "I have to do IVF" is true for a small group (no working tubes, severe male factor, no uterus, a serious heritable genetic disease) and oversold to everyone else. For most causes: unexplained infertility, PCOS, endometriosis, mild–moderate male factor, "low AMH" there are roads worth trying first.
1. The odds and why even the "official" numbers probably flatter reality
Even in the best-case group, one round of IVF fails more often than it works. Here is the industry's own US data next to the largest UK study so you can see the full range, not one cherry-picked figure.
|
|
SART (US, 2022) — live birth per intended retrieval |
McLernon 2016 (UK, 178,898 women) |
|---|---|---|
|
Under 35 |
43.1% |
~29.5% first cycle |
|
35–37 |
31% |
declines with age |
|
38–40 |
19% |
declines with age |
|
Over 42 |
3.2% |
declines with age |
|
Across ~6 cycles |
— |
~44% (conservative) to ~65–75% (optimistic) |
Per round, derived from McLernon's cumulative data (the chance each individual round works, for people still going): roughly 28% → 24% → 21% → 19% → 17% → 14% across rounds 1–6, the odds go down, not up, the more you do it. The cumulative total only climbs because you keep adding attempts, like buying more lottery tickets.
What both datasets agree on: a single round is a coin flip at best, it collapses with age (43% under 35 down to 3% over 42), and it doesn't improve with repetition. Even the highest, most flattering number ( 43% under 35 ) means 57% of retrievals in the youngest, healthiest patients don't end in a baby. Sources: SART, McLernon 2016.
And here's why even those numbers probably flatter reality. SART is a real, valuable source, reporting is federally mandated under the 1992 "Wyden law," and the CDC audits a random 7–10% of clinics each year. But:
-
It's the industry reporting on itself the professional society of the same clinics whose success it publishes, funded by member dues.
-
Only ~7–10% of clinics are audited each year; the rest is trusted self-entry. validation study
-
Harms are under-captured: birth outcomes are self-reported 8–9 months later, and researchers found the birth-defects field has poor sensitivity (it misses cases). validation study
-
SART membership is voluntary, so only member clinics report to it. source
-
Clinics can decline poor-prognosis patients, which raises reported success without falsifying a single number, the bias is in who gets counted. provider survey
-
ASRM (SART's parent) lobbies against additional oversight so the body reporting the data also has a policy stake. CBHD
-
It tracks success, not long-term safety. SART follows cycles → pregnancy → live birth. It does not follow mothers or children for years afterward. That data simply isn't in the registry by design.
This isn't a claim that SART is fake. It's the best data we have unfortunately. The point is narrower and fair: it's industry self-reported, only partly audited, weak at capturing harms, and silent on long-term outcomes, so even at its most flattering, it shows modest odds, and on the safety questions that matter most, we're probably not getting the full picture.
-
The headline number is often a pregnancy rate, not a baby. Clinics quote positive pregnancy tests (and "per transfer" instead of "per cycle started") because it sounds bigger than the take-home-baby rate. source
-
Clinics protect their stats by turning away hard cases. In one survey, 94% of fertility doctors said colleagues deny care to poor-prognosis patients to keep success rates high. PMC8581098
-
Before the 1992 reporting law, clinics advertised wildly inflated success rates which is exactly why mandatory reporting was created.
-
IVF "add-ons" are sold on hope, not evidence. The UK regulator rates 15 common add-ons: none green, 5 red (including PGT-A), 2 proven to do nothing. HFEA
-
Regulators have penalized fertility marketers for deceptive claims. FTC source
-
The clinic selling you the procedure also decides how many cycles and add-ons you "need." A structural conflict of interest.
-
A lot of people never needed it. In unexplained infertility, cumulative natural pregnancy reaches ~70–80% over 2–3 years in younger women (~45% over 35, ~30% after 5+ years). CFAS guideline, NCBI
-
Even after a failed IVF cycle, ~17–24% of couples conceive naturally. source
-
"Unexplained" often isn't, it's undiagnosed. In a 2024 study of 215 "unexplained" women (most post-failed IVF), over 90% had endometriosis once a surgeon looked. (Referral cohort higher than the general 30–63%.) Nezhat 2024
-
For many couples, monitored waiting matches IVF. source
-
ICSI is overused, ~70% of cycles, far above the rate of male infertility with no benefit (and possibly worse odds) absent a male factor. PET
-
Hormone stimulation: ~2 weeks of daily injections; ovaries swell 2–3×. ASRM
-
Egg retrieval is surgery: needle through the vaginal wall under sedation; bleeding, infection, ovarian torsion, rare organ injury. source
-
OHSS in ~1–5% of cycles, severe means chest/abdominal fluid, clots, kidney trouble, rarely death and even higher in those with PCOS. ASRM
-
Blood clots: IVF pregnancy carries ~4–10× first-trimester risk, up to ~100× after OHSS. source
-
Preterm birth higher even in singletons (~10% vs 5.5%), plus low birth weight. source
-
~30% of US IVF pregnancies are twins driving preterm, pre-eclampsia, NICU, cesarean. source
-
Pre-eclampsia ~2× with frozen transfer, ~4.5× with donor eggs. frozen, donor
-
Birth defects ~30–40% higher relative risk (higher with ICSI). meta-analysis
-
Neurodevelopment: ICSI linked to higher autism (RR ~1.5) and intellectual disability vs IVF; some studies show higher cerebral palsy (low-quality evidence but worth noting). source
-
80–85% of eggs never become a usable embryo. Source
-
Donor-egg pregnancies carry ~4.5× the risk of pre-eclampsia an immune mismatch between mother and baby (and no corpus luteum to regulate blood pressure). [PMC5844654]
-
IVF/ICSI children have a modestly higher risk of major birth defects: meta-analyses put it roughly 30–40% higher in relative terms (still a small minority of births). For IVF alone the rise is smaller and partly explained by the underlying infertility; ICSI carries the higher risk. [NEJMoa1008095; PMC9803464
-
Higher rates of ectopic pregnancy, placenta previa, and placenta problems and frozen transfers roughly double the risk of pre-eclampsia (8.6% vs 3.8%). [PMC7156607]]
-
Higher neurodevelopmental risk in the children. Compared with conventional IVF, ICSI is linked to higher autism (RR ~1.5) and intellectual disability; ART children overall show somewhat higher cerebral palsy rates than naturally conceived though much of that traces to prematurity. (Evidence is considered low quality) [PMC7734782]
-
The embryo's critical epigenetic programming happens in a lab dish (days 1–5); IVF is linked to rare imprinting disorders, and long-term cardiometabolic effects are still mixed/unknown. [PMC/Frontiers
-
80–85% of retrieved eggs never become a usable embryo. From ~10 eggs you might get 1–3. [CNY/Illume]]
-
PGT-A discards viable embryos, it tests placenta cells, not the baby; ~1 in 6 that led to healthy births would've been thrown out; rated red by the UK regulator. source
-
1–10 million frozen embryos in US storage; ~1.5–1.8 million unused yearly. source
-
The lab culture media isn't standardized - the brand can change the baby's birth weight. source
-
~4 in 10 women in treatment have anxiety or depression; 54–65% quit, mostly from the emotional toll; real strain on intimacy (but NOT increased divorce). mental health, divorce
-
Cost: ~$20k–$25k per cycle ($30k–$60k across multiple); most pay out of pocket. Source
-
The embryo's critical epigenetic programming happens in a lab dish (days 1–5); IVF is linked to rare imprinting disorders, and long-term cardiometabolic effects are still mixed/unknown. [PMC/Frontiers]
-
80–85% of retrieved eggs never become a usable embryo. From ~10 eggs you might get 1–3. [CNY/Illume]
-
IVF works around the disease, not on it. Endometriosis/adenomyosis are common and treatable & IVF treats neither and lowers your chance of a baby ~15% even after 3 rounds. Alson 2024
-
Same with PCOS: causes ~80% of ovulation-related infertility; ~half or more have insulin resistance IVF never addresses. Source
-
"Your BMI doesn't matter" but it does. Live birth drops ~7% per 5 BMI points; miscarriage rises. Source
-
"A TSH of 3.5 is normal" but under 2.5 is the preconception target (above ~4 is fairly linked to miscarriage; 2.5–4 is debated). Source
-
AMH is sold as a doomsday clock but it isn't. A poor predictor of natural conception; "low" AMH women still conceive (~65% within six cycles). Source
-
Standard testing is thin it often skips thyroid antibodies, full hormone panels, nutrient levels, imaging.
-
Chronic endometritis is a silent, treatable uterine infection (~14% of one "unexplained" group), usually fixed with antibiotics, not IVF. source
-
"We'll just suppress your endo with Lupron" isn't a fix is a temporary menopause; symptoms return. Excision surgery treats it. Source
-
The man barely gets looked at: infertility is male-factor ~half the time; varicocele repair improves sperm in ~70% of men. Source
-
In nature, the egg picks the sperm (progesterone "scent trail"; the body filters out weak, DNA-damaged sperm). ICSI throws that out. Source
-
Sperm chosen for ICSI more often carry DNA damage and epigenetic changes: the natural quality checks are skipped. Source
-
We don't fully know the long-term effects, we've only overridden natural selection since 1992; the first ICSI babies are barely in their 30s.
-
There are other roads worth trying first, restorative reproductive medicine treats the cause (26–41% live birth reported; observational, not proven head-to-head). Source
-
The US fertility industry is one of the least regulated in the developed world, most binding rules cover labs and reporting; embryo handling is largely voluntary guidance (the industry disputes this). CBHD
-
After Alabama clinics destroyed patients' embryos, the industry's response was a law shielding providers from lawsuits making negligence harder to sue over. PBS