How Accurate Is PGT Testing?

If you just got PGT results back, you probably want a straight answer: can I trust these numbers?

It depends which numbers you're looking at. If your embryo came back euploid or aneuploid, yes. Those calls are reliable. But if you got a mosaic result, the honest answer is that the classification itself is only right about half the time.

A 2025 meta-analysis of 109 studies found that 52.8% of mosaic calls turn out to be truly mosaic. That's barely better than a coin flip. And that gap between "the test reads DNA well" and "the test classifies your embryo correctly" is where most of the confusion lives.

Euploid and aneuploid calls work

When PGT-A calls an embryo euploid (normal chromosome count), you can trust that. The misdiagnosis rate after transferring a euploid embryo is 0.2%. If your embryo comes back normal, it almost certainly is.

Aneuploid calls are similarly strong. About 9 out of 10 embryos called aneuploid genuinely have a chromosome problem that would likely prevent a healthy pregnancy or cause a miscarriage. A separate meta-analysis put the overall sensitivity at 83.9%, and when you exclude mosaic cases, the accuracy jumps even higher. Not perfect, but good enough to genuinely reduce your chance of transferring an embryo that won't work.

Here's what the test actually does. PGT-A takes 5 to 10 cells from the trophectoderm, the outer shell of the embryo that becomes the placenta. It doesn't touch the inner cell mass, the part that actually becomes the baby. So the test is reading one tissue to make inferences about another. For clear results, this works well.

Some critics have pushed back harder. Gleicher et al. 2023 argued that a single trophectoderm biopsy is "mathematically unable" to determine embryo ploidy accurately enough for clinical use. And ASRM's 2024 committee opinion said that PGT-A as a universal screening tool is "unproven" for improving pregnancy rates. Both of those are real papers, but they're answering a different question. Whether every IVF patient should be screened is a question about clinical utility. Whether the test correctly identifies chromosomal status when it's used is a question about accuracy. The data on accuracy is strong.

The problems start with mosaics.

Mosaic results are unreliable

A mosaic classification means the biopsy found a mix of chromosomally normal and abnormal cells. That sounds precise. It isn't.

Only 35% of mosaic calls reproduce when the embryo is biopsied a second time. The other 65% come back with a different result entirely. And when four different labs tested identical samples, they got different euploidy rates. Part of this is biological (real mosaicism does exist in embryos), but a big part is that different labs use different thresholds. One lab calls an embryo mosaic at 20% variation from normal. Another uses 30%. The same embryo, sent to two labs, can get two different labels.

From the patient side, this is confusing. You get a mosaic result and you don't know whether a chromosome is actually missing or just flagged by the test. Nobody explains the difference between "the biopsy showed mosaicism" and "the embryo is mosaic."

But here's where it gets counterintuitive. The clinical outcomes for mosaics are much better than the classification accuracy would suggest. A prospective clinical trial found that low-degree mosaic embryos had a 42.9% live birth rate, compared to 43.4% for euploid embryos. No statistically significant difference. If your only option is a low-grade mosaic, the odds are better than you might fear.

That doesn't mean mosaics are identical to euploids. The largest study of mosaic transfers, covering 1,000 cases, found that mosaic embryos had a miscarriage rate of 20.4% compared to 8.6% for euploid transfers. Mosaics can work. But they carry real additional risk.

What matters most with a mosaic result is which chromosome is involved, the degree of mosaicism, and what your overall embryo cohort looks like. A mosaic on chromosome 16 is a very different conversation than a mosaic on chromosome 22.

PGT-P accuracy is a different question entirely

Everything above is about PGT-A, which counts chromosomes. PGT-P screens for disease risk using polygenic scores, and "accuracy" means something completely different.

The DNA readout itself is reliable. A study in Nature Medicine showed 99.0 to 99.4% genotyping accuracy at polygenic risk score sites from a day-5 biopsy. The test reads the DNA correctly.

The real question is whether those scores predict disease in real families. This is important because embryo screening is inherently a within-family decision. You're comparing embryos from the same two parents, not embryos from the general population. Siblings share roughly 50% of their genome, so the spread of scores between embryos from the same cycle is narrower than what you'd see across unrelated people.

A 2021 paper in the New England Journal of Medicine predicted that polygenic scores would lose predictive power in exactly this within-family context. The concern was reasonable. But it was tested empirically and, for disease traits, it was wrong. Moore et al. 2025 tested 17 disease scores on sibling pairs and found that 16 of 17 showed no decrease in predictive performance within families. For type 2 diabetes, families with high baseline risk could see risk reductions up to 23.5% by choosing among 10 embryos.

The science moved faster than the professional consensus predicted.

Whether this matters for you specifically depends on your embryo count, your family history, and which conditions concern you most.

What this means for your results

If your result is clear euploid or clear aneuploid, the test is doing its job. Trust those calls.

If your result is mosaic, understand that the label itself is unreliable, but the clinical outcomes for low-grade mosaics are better than the label suggests.

And if you're looking at PGT-P scores, the question to ask your provider is whether they validate on siblings. That's the only validation that matches what embryo screening actually does.

If you want help understanding what your PGT results mean, especially PGT-P scores where most genetic counselors lack deep expertise in polygenic scoring, our counselors can help. Please reach out to us.