Search "PCOS diet" in any forum and you'll find two camps within about three scrolls. Camp one: "keto saved my life, I reversed my PCOS, I've had two periods in a row for the first time in six years." Camp two: "keto destroyed me, I was exhausted for two months, I lost my period entirely, never again." Both camps are loud. Both camps are telling the truth.
This isn't a paradox. It's a mechanistic question: low-carb works via a specific pathway that is real and evidence-backed. But strict ketogenic eating — below 50g of carbohydrates per day — triggers a secondary pathway that, in a significant proportion of women with PCOS/PMOS, produces the opposite of what you're looking for. Whether you end up in camp one or camp two depends on which pathway dominates in your body, at your current hormonal baseline.
Atlas tells you which camp you're in. But first, you need to understand the four mechanisms.
Mechanism 1: Insulin and Androgen Suppression
This is the primary reason low-carb gets recommended for PCOS/PMOS, and the evidence is solid.
When you eat carbohydrates, your pancreas releases insulin to clear glucose from your bloodstream. In women with insulin resistance — which affects the majority of PCOS/PMOS cases — this insulin release is chronically elevated. High circulating insulin directly amplifies LH (luteinising hormone) pulse amplitude from the pituitary and signals the ovaries to produce more androgens: testosterone, androstenedione, DHEA. These elevated androgens drive the androgenic symptoms most associated with PCOS/PMOS: acne, hirsutism, irregular or absent cycles, the full cluster.
Carbohydrate restriction reduces insulin secretion. Less insulin means smaller LH pulses. Smaller LH pulses mean less ovarian androgen output. The insulin-androgen axis is the core mechanism, and it's why the dietary intervention works at all.
The evidence here is meaningful. The Mavropoulos 2005 pilot study (six months of a ketogenic diet in 11 women with PCOS/PMOS) found significant reductions in testosterone and LH:FSH ratio — and two of five women with previous infertility conceived during the study period. The Paoli 2020 review synthesised the growing body of evidence and found consistent androgen-lowering effects across low-carb interventions, independent of weight loss.
The nuance: this mechanism doesn't require ketosis. It requires insulin reduction. A moderate low-carb approach (50–100g carbs per day) produces meaningful insulin reduction. You don't need to go below 50g/day to access this benefit.
Mechanism 2: Blood Glucose Variability and the Glucose-Cortisol Feedback Loop
This mechanism is distinct from insulin sensitisation, and it's the one most PCOS/PCOS dietary advice glosses over.
Refined carbohydrates — white bread, sugar, processed snacks — don't just raise insulin. They raise it sharply, in a spike. That spike triggers a compensatory insulin surge that, in insulin-resistant bodies, often overshoots. Blood glucose drops below baseline: reactive hypoglycaemia, the "crash." The brain registers the drop as a starvation signal and triggers cortisol release to mobilise glucose from the liver via gluconeogenesis.
In a metabolically healthy person, this is mild and self-correcting. In PCOS/PMOS bodies — where baseline HPA axis activity is already dysregulated — this cortisol response stacks on top of an already-activated stress system. You get the crash, the cortisol pulse, the cravings surge, and the mood dip in sequence, repeating with every glucose spike across the day.
Low-carb eating flattens the glucose curve. Not by lowering overall glucose — by reducing variability. The sharp peaks and corresponding valleys become much smaller. The crash doesn't happen. Cortisol doesn't spike reactively. The afternoon cravings that follow a carb-heavy lunch at 1pm and hit you at 3pm? That's the glucose-cortisol loop playing out in real time. Flattening the glucose curve — through protein-first meals, reducing refined carbs, and moderating portion size of high-glycaemic foods — is one of the most direct ways to interrupt it.
This is worth understanding separately from insulin sensitisation because it explains why the dietary change is noticeable within days, not weeks. You don't need to wait for HOMA-IR to improve. You're just stopping the spike-crash-cortisol cycle, and you feel it almost immediately in your afternoon energy and cravings.
Mechanism 3: The HPA Axis Stress Response on Strict Keto — When It Goes Wrong
Here's the mechanism that explains the "keto ruined me" camp, and why it's not a failure of willpower.
Very low carbohydrate intake — below approximately 50g/day, the threshold for nutritional ketosis — can trigger a low-grade cortisol elevation in a significant proportion of women, particularly those with PCOS/PMOS. Three reasons:
First, the brain requires glucose to function. In ketosis, it adapts to run partially on ketones — but the transition takes 3–6 weeks, and during that window, the brain's remaining glucose demand must be met by gluconeogenesis: the liver manufacturing glucose from protein and fat. Gluconeogenesis is a cortisol-dependent process. More cortisol is needed simply to keep your brain fuelled. For women with PCOS/PMOS and already-elevated baseline cortisol, this additional load isn't trivial.
Second, the metabolic stress of keto adaptation itself activates the HPA axis transiently. This is sometimes described as "keto flu" — fatigue, brain fog, irritability in the first two weeks. In many women it resolves. In women with a primed stress response, it doesn't resolve — it becomes a chronic low-level elevation.
Third, and critically: cortisol is anti-gonadotropic. Elevated cortisol suppresses GnRH pulsatility, which disrupts LH/FSH signalling, which disrupts ovulation. This is the exact pathway that worsens cycle regularity. The same hormonal system that keto is supposed to help through insulin reduction can be undermined by the HPA axis stress response if you go too low.
The symptoms of this response are predictable: fatigue that doesn't improve after the first two weeks, sleep disruption (waking at 2–3am is a cortisol signal), worsening cravings particularly for sugar, and cycle irregularity. These are not detox symptoms. They are not a plateau. They are the mechanistic signal that strict keto is producing more HPA axis activation than your body can absorb.
The fix is not to push through. The fix is to ease back to moderate low-carb (50–100g/day) — retaining the insulin benefit without the gluconeogenesis load. Compare notes with intermittent fasting, which has a similar cortisol-dependent contraindication in PCOS/PMOS.
Mechanism 4: The Gut Microbiome and Fibre Trade-off
Keto dramatically reduces dietary fibre in most implementations. Most high-fibre foods — lentils, oats, wholegrains, legumes, many fruits — are carbohydrate-dense and typically eliminated on strict keto.
This matters for PCOS/PMOS specifically because the condition is increasingly linked to gut dysbiosis: an imbalance in the gut microbiome that amplifies both inflammation and insulin resistance. Fibre feeds the butyrate-producing bacteria (Bifidobacterium, Faecalibacterium prausnitzii) that reduce intestinal permeability and inflammation. Butyrate also stimulates L-cells in the gut to produce GLP-1 and PYY — hormones that improve insulin sensitivity and reduce appetite. The gut is a major input into the insulin-androgen axis, and fibre is the primary fuel for the microbiome species that support it.
A keto diet done without attention to fibre — relying primarily on animal products and eliminating vegetables — can inadvertently worsen the gut-hormone axis. You get the insulin benefit from reducing carbs while simultaneously undermining the microbiome that supports insulin sensitivity through a different pathway.
The practical solution: a well-formulated low-carb diet (50–100g/day) built around non-starchy vegetables, ground flaxseed, psyllium, and berries maintains adequate prebiotic fibre while keeping carbohydrates in the range that reduces insulin. You can also combine berberine with the dietary change for additive microbiome and insulin-sensitising effects. True keto (sub-50g) is harder to maintain gut health on — it requires deliberate supplementation with psyllium husk and non-starchy vegetables at every meal.
What the Research Actually Concludes
The evidence landscape here is genuinely nuanced, and you deserve the honest version.
The case for low-carb is strong. Multiple randomised controlled trials show improved insulin sensitivity, lower androgen levels, and better cycle regularity compared to standard Western diet — and these benefits appear without caloric restriction. The insulin-androgen axis mechanism is well-characterised. The glucose variability benefit is real and fast-acting. The research here is credible.
The case for strict keto is weaker. The evidence base for ketogenic diets specifically in PCOS is smaller, the trials are shorter-term, and critically — none of them tracked cortisol or thyroid function as primary outcomes. fT3 (active thyroid hormone) can suppress on very low carbohydrate intake because T4-to-T3 conversion is partially glucose-dependent. Neither the cortisol response nor the thyroid suppression appear in the headline results of keto-for-PCOS studies, but both are documented in other keto research. The absence from PCOS trials reflects gaps in study design, not absence of effect.
The practical conclusion: a moderate low-carb approach (50–100g carbs per day from vegetables, legumes in moderation, and berries) appears to capture most of the benefit of strict keto while avoiding the HPA axis and thyroid risks. Strict keto is a reasonable experimental step if moderate low-carb shows benefit — but it carries a clear contraindication signal: worsening fatigue, sleep disruption, or cravings at week 3–4 means the cortisol response is active. Ease back to moderate low-carb. That's information, not failure.
Five Atlas Tracking Signals for This Experiment
These five signals map directly to the mechanisms above. Log them daily throughout your 12-week protocol.
1. Morning energy score — your proxy for the overnight cortisol arc. On low-carb, morning energy should trend upward as the glucose-cortisol reactive cycle flattens. On strict keto, if the HPA axis cortisol response is occurring, morning energy will worsen — particularly in weeks 2–4. This is the earliest signal and the most diagnostic.
2. Cravings severity (afternoon and evening) — your proxy for glucose variability. As refined carbs reduce and the glucose curve flattens, the 3pm craving surge should soften within days. If cravings remain high or worsen after week 2 of a carb reduction, look at whether you've inadvertently replaced carbs with very low protein — undereating protein re-activates the brain's fuel-shortage signal.
3. Mood and cognitive clarity — your proxy for HPA axis stability. Moderate low-carb typically improves both within 2–3 weeks as the glucose-cortisol loop quietens. Strict keto can temporarily worsen both during adaptation, and if they haven't recovered by week 4, that's the cortisol signal, not a plateau.
4. Skin and acne flag — your proxy for androgen suppression. This is the slowest-moving signal: androgen reduction takes 6–12 weeks to manifest as visible skin change. Don't expect week-2 results. Log it anyway — the 12-week comparison is what reveals the trend.
5. Cycle length variance — your proxy for LH pulse normalisation. Cycle changes take longest of all: expect 12+ weeks before they register. Some women see cycle length shorten meaningfully in the first three months; others need six. The signal is real, but patience is required. Log your cycle day alongside your daily check-in and let Atlas plot the trend.
The 12-Week Protocol
Experiment 1 — Weeks 1–4: Baseline
Don't change your diet yet. Log your current morning energy, afternoon cravings, mood, skin condition, and cycle day every day. This is not wasted time — it's the comparator that tells you whether the intervention actually worked. Without a baseline, you're guessing at the direction of change.
Note particularly: what does your current craving pattern look like? Is there a reliable afternoon low? What does your morning energy score average? What's your current cycle length? These numbers become your personal benchmark.
Experiment 2 — Weeks 5–8: Moderate Low-Carb (50–100g/day)
Remove refined carbohydrates and added sugar. Prioritise protein at every meal — aim for 30g at breakfast, protein-forward lunch and dinner. Keep non-starchy vegetables, berries, and legumes in moderation. Do not go below 50g of total carbohydrates at this stage.
What you're targeting: the insulin suppression and glucose variability mechanisms, without approaching the cortisol threshold that strict keto can trigger.
Track your five Atlas signals against your baseline. By week 7–8, you should see: reduced afternoon cravings, improved morning energy, and stabilised mood. If none of these shift, the insulin-androgen pathway may not be the primary driver for your presentation — the experiment has still been valuable.
Experiment 3 — Weeks 9–12: Optional Strict Trial or Consolidation
If weeks 5–8 showed clear improvement in energy, mood, and cravings with no new symptoms, you have a choice: consolidate at moderate low-carb (the sustainable default for most women), or trial going below 50g/day to test for additional benefit.
If you trial strict: watch morning energy, mood, and cravings specifically in weeks 10–11. If any of the three worsens versus your weeks 5–8 average, that is the HPA axis signal. Return to moderate low-carb immediately. You haven't failed the experiment — you've found your personal threshold, which is the most useful thing this protocol can tell you.
If you don't trial strict: stay at moderate low-carb and use weeks 9–12 to let the skin and cycle signals develop. These are the signals that require the most time.
What to Ask Your GP
Dietary interventions for insulin resistance have specific measurable outputs. Ask for these:
- Fasting insulin and HOMA-IR (not just fasting glucose — glucose alone misses insulin resistance in its early stages)
- HbA1c at 12 weeks — the 3-month average glucose, a direct measure of whether your dietary change has moved the needle
- Thyroid panel (TSH, fT3) at 12 weeks if trialling strict keto — fT3 can suppress on sub-50g carbohydrates; this is rarely tracked in standard care but matters if you feel persistently cold, fatigued, or cognitively sluggish
These numbers tell you whether the diet is working at the biochemical level, not just symptomatically. If HOMA-IR drops from 3.5 to 2.1 over 12 weeks, the intervention is working regardless of what the scale says.
Both Camps Are Right — For Different Carb Levels
The keto evangelists and the "keto ruined me" accounts aren't contradicting each other. They're describing the same intervention at different carbohydrate thresholds, in bodies with different HPA axis baselines.
Moderate low-carb (50–100g/day) reduces insulin, flattens glucose variability, and supports the gut-hormone axis — the three mechanisms with the strongest evidence. Strict keto adds a fourth potential benefit (deeper insulin suppression) but also adds gluconeogenesis cortisol load and fibre depletion risk. Whether the fourth benefit outweighs the additional risks depends entirely on your individual cortisol baseline and HPA axis sensitivity.
Atlas tells you which. Not by guessing at your hormonal profile — by showing you whether morning energy, mood, and cravings move in the right direction or the wrong one as you cross the 50g threshold.
Track your energy, cravings, skin, and cycle alongside your dietary experiments in Atlas. The pattern engine surfaces your personal response — not the average.