Metformin and Fertility in PCOS: Where It Helps and Where It Does Not

Does metformin help you ovulate?

Metformin raises the odds of ovulating 2.64-fold over placebo in women with PCOS (95% CI 1.85–3.75), according to a 2019 Cochrane review pooling 41 randomised trials in 4,552 women on metformin versus placebo. That is a real, consistent effect, and it is the reason metformin appears in fertility conversations at all.

The mechanism is indirect. Metformin lowers circulating insulin, and high insulin is one of the drivers of the ovarian androgen excess that stops follicles maturing in PCOS. Bring insulin down and some women resume regular ovulation on that basis alone — which is also why the effect is least reliable in women whose PCOS is not driven primarily by insulin resistance. For the full mechanism and side-effect picture, metformin for PCOS covers the drug itself; this page is about what that mechanism does and does not do for a pregnancy.

Ovulating is not the same as getting pregnant

The same Cochrane review found metformin raised the odds of clinical pregnancy by roughly 2-fold (OR 1.98, 95% CI 1.47–2.65) — smaller than the ovulation effect, and a signal that something is lost between ovulating and conceiving. The live-birth result was smaller again and less certain: OR 1.59 (95% CI 1.00–2.51), with live birth occurring in about 19% of women on placebo and 19–37% of women on metformin depending on the trial.

That gap matters. Ovulation is necessary for pregnancy but not sufficient — egg quality, tubal patency, endometrial receptivity and a partner’s sperm parameters all sit downstream of it, and metformin does not touch any of them. A drug that reliably restores ovulation can still produce a disappointing live-birth number if those other factors are also in play.

Metformin to get pregnant: the head-to-head numbers

The clearest single data point on metformin’s fertility performance comes from a 626-woman, multicentre randomised trial that compared clomiphene alone, metformin alone, and the combination. Live-birth rates were 22.5% with clomiphene, 7.2% with metformin alone, and 26.8% with both. Metformin alone finished a distant third. Adding it to clomiphene produced a small numerical gain over clomiphene by itself, but the trial and the later Cochrane pooling both treated that gain as uncertain rather than proven.

Why the guideline’s first choice is letrozole, not metformin

The 2023 international guideline names letrozole, not metformin, as the first-line drug for ovulation induction in PCOS. In a separate 750-woman trial, women assigned letrozole had more cumulative live births than those on clomiphene — 103 of 374 (27.5%) versus 72 of 376 (19.1%), rate ratio 1.44, 95% CI 1.10–1.87 — with a higher ovulation rate to match (61.7% of cycles versus 48.3%). Metformin was not part of that comparison at all; letrozole earned first-line status against clomiphene, the older standard, not against metformin. The full letrozole-versus-clomiphene picture, including side effects and why the guideline changed its recommendation, is in Clomid vs Letrozole for PCOS.

Where metformin’s fertility role actually sits

None of this makes metformin fertility-irrelevant — it makes it a second-tier tool rather than a starting point. The guideline discusses combination approaches, including adding metformin, as one option a fertility specialist may weigh for women who do not ovulate on first-line treatment alone or who have significant insulin resistance alongside anovulation. That is a decision made in an appointment, against your specific labs and history, not a sequence to arrange yourself. If you have not worked out whether insulin resistance is part of your picture, can you get pregnant with PCOS covers what actually predicts conception chances beyond ovulation status alone.

One finding from the Cochrane data is worth naming even though it settles nothing: when trials compared metformin head-to-head with clomiphene, results split by body size. Women in the higher-BMI trials tended to do worse on metformin relative to clomiphene, while the trials restricted to non-obese women leaned the other way. The reviewers rated this evidence very low quality and explicitly declined to draw a conclusion from it. It is one plausible reason a specialist might weigh metformin differently depending on your starting BMI — not a rule, and not something to act on without that appointment.

Metformin for cycles versus metformin for fertility

Many people are prescribed metformin to turn a cycle that runs to 40, 60 or 90 days into something closer to predictable, with no active pregnancy attempt underway at all — and that is a different goal from getting pregnant, backed by different evidence. Getting periods to arrive more often is not the same claim as getting pregnant, because a bleed does not confirm ovulation occurred, and the fertility trials above measured ovulation and pregnancy directly rather than cycle timing. If cycle regularity, not conception, is your actual question, that is a different set of trials than the ones summarised here.

Metformin during pregnancy: what the evidence shows

A 487-woman trial across 14 hospitals in Norway, Sweden and Iceland — PregMet2 — randomised women with PCOS to metformin or placebo, taken through pregnancy, to test whether it reduced a combined outcome of late miscarriage and preterm birth. That is a separate question from getting pregnant: women with PCOS enter pregnancy with a higher baseline rate of complications than the general obstetric population, which is the reason this question has been trialled at all — the detail is in PCOS in pregnancy. In the trial’s own result, the composite outcome occurred in 5% of the metformin group versus 10% of the placebo group — a promising direction, but the confidence interval (OR 0.50, 95% CI 0.22–1.08) did not reach statistical significance on its own. Gestational diabetes was essentially unchanged between groups (25% versus 24%), and the trial reported no excess of serious adverse events attributed to the drug in either mothers or babies.

A post-hoc analysis pooling this trial with two earlier, smaller ones found the reduction in combined late miscarriage and preterm birth did reach significance across the pooled data — but a post-hoc pooled analysis is weaker evidence than a single trial designed to answer the question, and the guideline has not elevated this to a routine recommendation. PCOS in pregnancy covers what is actually known about these risks and what monitoring exists for them regardless of what medication you are on.

What this will not do for you

• It will not outperform letrozole for ovulation induction. The guideline evidence runs the other way.
• It will not fix causes of infertility outside the ovulation pathway. Tubal blockage, male-factor infertility and endometrial issues are untouched by an insulin-sensitising drug.
• It has not been proven to prevent pregnancy loss. PregMet2’s own primary result did not reach significance; only a secondary, pooled analysis did.
• It works less predictably if insulin resistance is not your main driver. The mechanism has less to act on, which is likely why average effect sizes across trials are modest rather than dramatic.
• It does not act quickly. Every trial above measured its outcomes across three to six treatment cycles — call it three to six months as a realistic window, not weeks.

You may see PCOS written as polyendocrine metabolic ovarian syndrome (PMOS), after a 2026 global consensus of more than 50 organisations renamed it. Every trial above still applies under either name — only the label changed.