How Long Does Metformin Take to Work for PCOS (Now PMOS)? A Timeline, Outcome by Outcome

You were handed a prescription and a sentence: give it a few months. Nobody said a few months of what. A few months before your period arrives on time? Before the weight shifts? Before it's fair to conclude the whole thing isn't working?

The reason that sentence is so vague is that there isn't one answer. Metformin doesn't improve every part of PMOS at the same rate — and the studies that measured each outcome separately found genuinely different clocks. Some things move in weeks. One thing moves at two months and then quietly gives the ground back. Others barely move at all.

PCOS is now PMOS — Polyendocrine Metabolic Ovarian Syndrome, renamed by international consensus in May 2026. Both names are still in circulation, so both appear here. Nothing below is medical advice, and your dose and duration are decisions for you and whoever prescribed it.

The short answer, by outcome

What you're watching	When it's fair to judge
Blood sugar and insulin	Days to weeks — but you won't feel it
Menstrual regularity	3 months at the earliest; 6 months is the real checkpoint
Ovulation	3–6 months, and it tracks with cycle regularity
Androgen symptoms (skin)	Slow, partial, and not guaranteed
Hirsutism (unwanted hair)	Often not at all — this isn't what metformin is for
Weight	Modest at best; not the primary mechanism

The single most common mistake is judging metformin at eight or twelve weeks on the outcome that takes the longest.

Blood sugar moves first, and invisibly

Metformin works on the liver and on how your tissues handle insulin, and that biochemistry starts changing early — well before anything shows up in your calendar or your mirror.

This is the part that misleads people. Nothing about the first month feels like the drug is working. What you're likely to notice instead is the gastrointestinal adjustment: nausea, loose stools, a stomach that objects to breakfast. In the Cochrane review of insulin-sensitising drugs, gastrointestinal side effects were nearly five times more common on metformin than placebo (OR 4.76) — comfortably the most reliable short-term effect in the entire evidence base.

So the honest framing for month one is: the side effects arrive before the benefits, and that ordering is normal rather than a signal that it's the wrong drug.

Cycles: three months is too early, six months is the checkpoint

This is where the useful number lives, and it comes from a study that specifically compared women treated for different lengths of time.

Researchers at Virginia Commonwealth University reviewed women with PCOS on metformin, splitting them into those followed for 3–6 months and those followed for longer than 6 months. Overall, 69% responded with improved menstrual cyclicity, and 88% of those responders reached normal cyclicity. But the split by duration is the finding worth carrying: 77% responded in the group treated beyond six months, versus 55% at 3–6 months.

Read that carefully, because it changes what "it isn't working" means. A woman who stops at four months with irregular cycles has not learned that metformin failed her. She has landed in the group where roughly one in two responds. The same woman at seven months would be in the group where roughly three in four do.

This was a retrospective review rather than a randomised trial, so it describes what happened rather than proving the extra months caused it. But it is the most direct duration comparison available, and it points the same way as the pooled evidence: across seven trials in 427 women, metformin improved menstrual frequency against placebo (OR 1.72), and across fourteen trials in 701 women it improved ovulation (OR 2.55).

If you're tracking anything, track cycle length month to month rather than waiting for one perfect 28-day cycle. Cycles that shorten from 90 days to 50 are responding, even though nothing about them looks normal yet.

The androgen dip that doesn't hold

Here is the finding that almost nobody writes about, and it's the reason a timeline by outcome is worth more than a single number.

A prospective study of 20 women with PMOS, treated with 500 mg three times daily for four to six months, measured hormones throughout. Total testosterone dropped significantly at two months — and had returned to its starting level by four to six months. Free testosterone did decrease and stayed down. Cycles improved in 11 women, 68.8% of those who had menstrual disturbance to begin with.

So the picture is mixed rather than tidy: the bound-and-free androgen story doesn't move as one block, and the earliest hormonal change was also the least durable. It's a small study from 1998 in obese participants only, so it isn't the last word. But it is a measured account of something people experience and then can't explain — a few good months on skin, followed by a plateau.

Hair and weight: the outcomes metformin is worst at

In that same study, over four to six months, there was no change in hirsutism, no change in body mass index, and no change in blood pressure.

That is not a failure of the drug. It's a mismatch between what metformin does and what it's often quietly expected to do. Unwanted hair follows a follicle cycle that runs on its own schedule and responds to anti-androgen strategies rather than insulin-sensitising ones — which is why spironolactone's timeline looks completely different, and why hair is the outcome where impatience is most misplaced.

Weight is the other one. Metformin can help modestly for some people, and it is positioned in the 2023 International Evidence-Based Guideline around metabolic features rather than as a weight-loss drug. If weight is the only thing you're measuring, you're likely to conclude it isn't working while the thing it is doing goes unnoticed.

If you'd like a clearer read on which driver is actually loudest for you — insulin resistance, androgens, ovulation, SHBG or cortisol — our free assessment takes a few minutes and returns a driver-level picture rather than a diagnosis.

What "it isn't working" honestly means, and when to say it

Three things worth having straight before you make that call:

Give it six months, not three, if cycles are the target. The duration data is unusually clear on this, and stopping early is the most common way to get a false negative.

Be specific about which outcome failed. "It didn't work" usually means "my weight didn't change" or "my hair didn't change" — neither of which metformin was ever likely to fix. That is different from six months at a reasonable dose with no movement in cycle length at all.

Distinguish side effects from failure. If the gastrointestinal effects never settled, that's a formulation and dosing conversation — extended-release exists precisely for this — rather than evidence that the mechanism doesn't suit you.

And if the question underneath all of this is really how long am I going to be on this — that's a different question with a different answer, because what comes back when you stop depends on how long you were taking it. If you're weighing metformin against the supplement route, we've compared them directly here.

Sources

• Essah PA, Apridonidze T, Iuorno MJ, Nestler JE. "Effects of short-term and long-term metformin treatment on menstrual cyclicity in women with polycystic ovary syndrome." Fertility and Sterility 2006 Jul;86(1):230–2. PMID 16716324. (Retrospective review; response 69% overall, 77% at >6 months vs 55% at 3–6 months.)
• Morin-Papunen LC, Koivunen RM, Ruokonen A, Martikainen HK. "Metformin therapy improves the menstrual pattern with minimal endocrine and metabolic effects in women with polycystic ovary syndrome." Fertility and Sterility 1998 Apr;69(4):691–6. PMID 9548159. (Prospective, n=20 obese participants, 4–6 months; testosterone fell at 2 months and returned to baseline by 4–6 months; no change in hirsutism, BMI or blood pressure.)
• Morley LC, Tang T, Yasmin E, Norman RJ, Balen AH. "Insulin-sensitising drugs (metformin, rosiglitazone, pioglitazone, D-chiro-inositol) for women with polycystic ovary syndrome, oligo amenorrhoea and subfertility." Cochrane Database of Systematic Reviews 2017 Nov 29;11(11):CD003053. PMID 29183107. (48 studies, 4,451 women; metformin vs placebo — ovulation OR 2.55, menstrual frequency OR 1.72, gastrointestinal side effects OR 4.76.)
• Kravos NA, Janež A, Goričar K, Dolžan V, Jensterle M. "Effects of metformin withdrawal after long and short term treatment in PCOS: observational longitudinal study." Diabetology & Metabolic Syndrome 2021 Apr 12;13(1):43. PMID 33845893. (Observational, n=40, no control group.)
• Teede HJ, et al. International Evidence-Based Guideline for the Assessment and Management of Polycystic Ovary Syndrome, 2023.