These three come up constantly and get talked about as though they were rival options — as if there were a best one and the job is to find out which. That framing is the confusion. They are aimed at different problems, and the sensible question is not “which is best” but “what is being treated here”.
This page names no doses and recommends nothing. What follows is what each is for, so that a prescribing conversation is one you can follow rather than one that happens to you. What you should take is a decision for you and a clinician who knows your history.
The combined oral contraceptive pill
What it is aimed at: menstrual irregularity and the symptoms of raised androgens.
The 2023 international guideline names it as first-line pharmacological management for both. It does not name a particular preparation as superior, and states a preference for lower ethinyl estradiol doses and formulations with fewer side effects — which is worth knowing, because it means there is legitimate room to change if a given one does not suit you. “The pill did not agree with me” is information, not the end of the option.
What it does not do: it does not cure PCOS, and it does not fix ovulation — it replaces the cycle rather than restoring it. The bleed on a combined pill is a withdrawal bleed, not a period following ovulation. People are often surprised, on stopping it after several years, that the irregularity is exactly where they left it. That is not the pill having “hidden” anything or “caused” PCOS; it is the underlying pattern reappearing once the replacement stops.
What it is often prescribed for besides: endometrial protection. Prolonged gaps between periods mean the endometrium is not shed regularly, and that has its own long-term considerations. If nobody has explained why regular bleeding matters in itself, this is usually the reason — and it is a good thing to ask about directly.
Metformin
What it is aimed at: the metabolic side. Insulin resistance, glucose and lipid levels.
The guideline positions metformin primarily for metabolic features, and specifically recommends it for adults with a BMI at or above 25 mainly for that purpose. It is also the comparison against which supplements get measured: the same guideline concludes that metformin has greater efficacy than inositol, which offers limited clinical benefit.
What it is not: a weight-loss drug, although it is frequently discussed as one. Any weight effect is modest and is not what it is prescribed for here.
The practical part nobody mentions until it happens: gastrointestinal side effects are common at the start and are the usual reason people stop. If that is what happened to you, it is worth saying so rather than filing it under “did not work” — there are formulations and approaches to starting it that exist precisely because of this, and a clinician can tell you which apply.
Spironolactone and other anti-androgens
What it is aimed at: the visible androgen symptoms — hirsutism in particular, and sometimes scalp hair thinning.
The guideline gives anti-androgens a limited role, for situations where other therapies have been ineffective or are contraindicated. That is a more restricted position than their popularity in online discussion suggests, and the reason is straightforward: the evidence base is thinner than for the first-line options.
Two things a clinician will raise with you, and should:
Prescribing involves contraception considerations, which is why anti-androgens and the combined pill are often discussed together rather than as alternatives.
And the timescale is long. Hair growth cycles mean any change in hirsutism takes months to become visible — several months at minimum before an effect can be assessed at all. Stopping at six weeks because “nothing happened” is stopping before the question could have been answered.
Why “which is best” has no answer
Line them up by what they address:
| Aimed at | Guideline position | |
|---|---|---|
| Combined pill | Menstrual irregularity, androgen symptoms | First-line for both |
| Metformin | Metabolic features — insulin, glucose, lipids | Primarily metabolic; recommended in adults with BMI ≥ 25 for that purpose |
| Anti-androgens | Visible androgen symptoms | Limited role, where other therapies are ineffective or contraindicated |
Nothing in that table competes with anything else in it. Someone whose main problem is 60-day cycles and someone whose main problem is a fasting glucose creeping upwards are having two different conversations, and the fact that both have PCOS does not make them one conversation.
Which is why the useful preparation for a medication appointment is not researching drugs. It is being able to say precisely what the problem is: how many periods since when, which symptoms on how many days, what the last set of bloods said and when.
The question that decides whether it is working
Whatever is started, ask two things before you leave:
What should change, and by when? What would make us stop or switch?
Both are answerable, and the answers give you something to measure against. Without them, three months later the assessment is “I think maybe a bit?”, which is not enough to act on and is heavily shaped by what you hoped for.
With a record of the months before and the months after — cycle dates, symptom frequency, the labs and their dates — the assessment is two sets of numbers. That comparison is also the only thing that reliably separates a real effect from the ordinary variation of a condition that comes and goes on its own.
What is not on this page, and why
No doses, for anything, including the things sold without a prescription. A dose depends on you: what else you take, what your kidneys are doing, what you are being treated for. Publishing one is treatment advice, and it is advice given without any of that information.
No ranking, either. The guideline’s positions above are about populations. Yours is a conversation with someone who knows your history — and the more of that history you can put in front of them, the better that conversation goes.