Masters & Female General Endurance · · 10 min read

Training by Menstrual Cycle Phases: What the 2026 Research Actually Says (and What It Doesn't)

The 'lift heavy in follicular, deload in luteal' rule isn't evidence-based. When McNulty 2020 filtered to high-quality studies only, the effect size was -0.01 — functionally zero.

AO
AthleteOS Data Science
TL;DR — The Answer

The McNulty 2020 meta-analysis (78 studies, 1,193 participants) found a pooled effect size of -0.06 — trivial — collapsing to -0.01 in the 16 highest-quality studies: no RCT supports the popular cycle-phase prescription. The 2024 Hammes RCT had athletes train deliberately against conventional guidance and they gained the same VO2max (39 to 43 mL/kg/min) as those following it. What IS supported: luteal-phase heat strain, HRV shifts, iron status, and N-of-1 personal logging.

The popular prescription sounds convincing: go hard in the follicular phase, back off in the luteal. Filter the data to the 16 highest-quality studies in the biggest meta-analysis ever done on this topic, and the effect size is -0.01 — one step above zero. That’s the gap between what the fitness industry is selling and what the research shows. There are real, concrete things you can track instead.


TL;DR


What the 2020 Meta-Analysis Actually Found About Menstrual Cycle Phase

The McNulty et al. 2020 meta-analysis is the largest study of its kind: 78 studies, 1,193 participants. The researchers asked whether exercise performance differs between menstrual cycle phases.

The pooled effect size across all comparisons was ES = -0.06. That’s trivial. Anything below 0.2 in exercise science is considered too small to matter in practice.

Then they filtered down to the 16 studies rated highest quality. Those showed an effect size of ES = -0.01. Basically zero.

Why the difference? Because 66% of the studies in the full dataset were rated “low” or “very low” quality. When you strip out the noisy data, the already-thin signal disappears entirely.

The authors’ own conclusion: a personalized approach, based on each athlete’s individual response, is more supported than any population-level prescription.

Here’s how the effect sizes sit relative to the threshold that would make them meaningful.

Menstrual Cycle Phase vs Performance: Effect Sizes (McNulty 2020) Trivial threshold (|ES| = 0.20) 0.20 threshold Late Follicular vs Early Follicular ES -0.14 Ovulation vs Early Follicular ES -0.07 Early Luteal vs Early Follicular ES -0.07 Mid-Luteal vs Early Follicular ES -0.04 Late Luteal vs Early Follicular ES -0.01 Overall pooled (all studies) ES -0.06 High-quality studies only (n=16) ES -0.01 All values fall below the trivial threshold of 0.20. High-quality studies show ES = -0.01. Source: McNulty et al., Sports Medicine 2020.

Every single comparison is below the trivial threshold. The “worst phase” is barely distinguishable from the “best phase.”

Your cycle phase doesn’t predict your performance the way the apps claim.

The 2024 RCT Nobody Talks About

If you want to test whether cycle-phase training works, you need to randomize athletes to train against the prescription and compare them to athletes following it.

Hammes et al. 2024 did exactly that. Thirty-three naturally menstruating, moderately trained women. Eight weeks of polarized training. Two groups:

Both groups improved from roughly 39 to 43 mL/kg/min VO2max. No significant group-by-time interaction on any parameter.

Translation: the athletes who deliberately trained against the conventional prescription got the same fitness gain.

That’s not a null finding buried in noise. It’s a direct test of the core claim, and the core claim failed.

What the Menstrual Cycle Does Affect (and How Much)

The point isn’t that your cycle doesn’t matter. It does. Just not in the way the prescription apps suggest.

FactorWhat the data showsPractical implication
Heat strain (luteal)Resting core temp is 0.3 to 0.6°C higher post-ovulation. JSAMS 2020 found pooled ES = 1.23 for greater initial body temp in luteal vs follicular.Pre-cool; pace conservatively in late-luteal heat.
HRV shift37-study meta-analysis: follicular-to-luteal HRV decrease of d = -0.39 (medium effect). Average HRV drop of 3.2%, resting heart rate up 1.6%.Log HRV against cycle phase; don’t panic at the drop.
PMS / late-luteal symptoms78% of athletes in a 1,086-person survey reported PMS. But meta-analysis: PMS prevalence 25.7% in athletes vs 30.1% in non-athletes — no significant difference.Use RPE on the day, not a fixed luteal template.
Iron status46% of competitive women iron deficient (ferritin below 30 µg/L). Menstruation is the primary modifiable iron-loss variable.Test ferritin annually; don’t wait for symptoms.
Plasma volumeFollicular-to-luteal change of just -2.4%; all comparisons non-significant (p = 0.48 to 0.93).Often cited as a performance mechanism; the data say it’s negligible.
Bone stress and REDs2023 IOC REDs consensus: REDs indicators in 23 to 80% of female endurance athletes.Cycle irregularity signals underfueling, not a training variable.

Heat in the luteal phase is the most actionable finding. Think of it like this: your thermostat is set 0.5°C higher for two weeks out of every four. On a cool day, that barely matters. In a summer race, you’re starting every mile slightly deeper in the heat hole than you were last week. That’s worth planning around.

Why the Tracking Apps You’re Using Are Probably Wrong

Cycle-phase training requires knowing which phase you’re actually in. That means accurately detecting ovulation.

Johnson, Marriott and Zinaman (2018) tested 949 women and compared app-based ovulation prediction against LH surge testing. App accuracy: no better than 21%.

Calendar-based apps are wrong about ovulation more than three-quarters of the time. If you don’t know which phase you’re in, cycle-phase prescriptions can’t work — regardless of whether the underlying biology holds up.

Here’s how the main tracking methods compare.

MethodOvulation accuracyCostEffortBest for
Calendar app (Clue, Flo)≤21%FreeLowRough cycle log only
Basal body temp (BBT)~75% retrospectiveLowDaily AM measureConfirming ovulation occurred
LH urine strips>95%~$25/cycleDaily ~7 days midcycleReal-time ovulation detection
Continuous core temp ringHighHighPassiveAthletes already wearing one

Source: Johnson et al. 2018; BBT accuracy from standard rhythm-method literature.

LH urine strips tested daily around your expected ovulation window give more reliable phase identification. BBT works but tells you after the fact. Neither is built into most training apps.

The Hormonal Contraceptive Problem

Around 50 to 63% of female athletes use hormonal contraceptives, depending on the population surveyed. A 2024 cross-sectional study of 323 female endurance athletes found 51% used hormonal IUDs and 29% used oral contraceptive pills.

Natural-cycle prescriptions don’t apply to this group. HC users don’t cycle through the same estrogen and progesterone fluctuations.

What does the evidence say about HC and performance? The Elliott-Sale et al. 2020 meta-analysis (42 studies, 590 participants) found the effect of oral contraceptives on performance was trivially small: between-group effect size 0.13 to 0.18, within-group 0.05. HC doesn’t tank your performance — it makes follicular/luteal prescriptions irrelevant.

If you use HC, the more useful tracking anchor is pill phase: active pill week vs inactive or placebo week. Some athletes notice lower HRV or heavier fatigue during the inactive week. That’s worth logging. Don’t map your experience onto a natural-cycle framework.

Why the N-of-1 Playbook Beats Population Prescriptions

Take Priya, a 34-year-old marathon runner targeting a 3:20. She’d spent two months following a cycle-sync app. The app scheduled hard sessions during “peak follicular.” Her long-run RPE was all over the place — some “luteal” sessions felt great, some “follicular” ones were brutal.

When she started logging her own HRV, RPE, and cycle phase daily, 12 weeks of data told a different story. Her HRV dropped sharply in the 48 hours before her period started — not in mid-luteal as the app assumed. Her hardest sessions consistently landed hardest in that window. Now she shifts any key workout landing there by 48 hours. One simple adjustment, grounded in her own data.

Between-study variation in the McNulty meta-analysis was large (tau = 0.26) relative to the pooled effect (ES = -0.06). Some athletes do show real phase patterns. You can’t tell from population data whether you’re one of them.

Here’s a protocol that works this month.

  1. Log cycle start dates. Calendar-based phase ID is imprecise. Add LH strips around day 10 to 14 if phase accuracy matters.
  2. Log readiness daily. Resting HRV, sleep quality, RPE at a fixed submaximal pace.
  3. Run 3 full cycles before drawing conclusions. One cycle is noise. Three start to show signal.
  4. Look for your own repeating patterns. Not the population average — your pattern.
  5. Adjust intensity on the day when symptoms warrant — not on a phase template.
  6. Get ferritin checked annually. Twice yearly for heavy bleeders. Iron deficiency explains fatigue that looks like a luteal slump.

For more on reading HRV without overreacting to daily noise, see HRV readiness trend: daily noise vs 7-day signal.

How AthleteOS Handles Menstrual Cycle Data

AthleteOS takes the N-of-1 approach rather than imposing a population template. Athletes can log cycle phase on every session entry. AthleteOS then overlays that phase data against HRV trends, session RPE, and fitness score patterns across your own history.

After 3 or more cycles, AthleteOS surfaces your personal correlations. If your HRV consistently drops 6% in late luteal, you’ll see that in your dashboard. If it doesn’t, you’ll see that too. The AI coach doesn’t prescribe based on what phase you’re in — it uses your data to surface whether your phase actually predicts your readiness.

For HC users, there’s a toggle that switches from natural-phase labels to pill-phase labels (active vs inactive week).

This connects directly to what’s covered in training past 50: how masters athletes adapt, where personal response variability is an equally central theme.

If you’re unsure whether your fatigue is hormonal or nutritional, see the AthleteOS guide on sodium and heavy sweaters for a parallel framework on tracking your own metabolic patterns.

To start logging your cycle phase alongside training data, connect your workouts in AthleteOS and enable the menstrual cycle tracking field in your health settings.


The research doesn’t say your cycle doesn’t matter. It says the population average is too small to build a training plan around. Your individual pattern might be real, meaningful, and worth acting on.

The only way to find out is to log your own data long enough to see it.

Start logging your cycle phase in AthleteOS and let your own numbers tell you what the research can’t.

Frequently Asked Questions

Should I deload in the luteal phase?

There's no RCT evidence that a blanket luteal deload improves performance. The 2024 Hammes RCT found athletes who trained hard in the luteal phase gained the same VO2max as those who took it easy. If PMS symptoms are genuinely limiting you on a given day, modify that session. Don't modify the whole phase by default.

Are cycle-tracking apps accurate for training purposes?

No. Johnson et al. 2018 tested 949 women and found app-based ovulation prediction is no better than 21% accurate. Since cycle-based training requires knowing your actual phase, calendar-only apps fail before the training question even matters. LH strips or basal body temperature give more reliable phase identification.

What if I'm on hormonal contraceptives?

Cycle-phase prescriptions don't apply to you. HC use among female athletes runs 50 to 63% across surveys. The OCP effect on performance is also trivial (between-group effect size 0.13 to 0.18). Track your symptoms and HRV against pill phase instead — active vs inactive week is more physiologically relevant.

Does my HRV always drop in the luteal phase?

The average drop is about 3.2%, with a medium effect size (d = -0.39 across 37 studies). But individual variation is large. Some athletes see no drop; others see a consistent 8 to 10% suppression. Log HRV and cycle phase together for at least 3 cycles before drawing any conclusions about your own pattern.

What about heat races during late luteal?

This is where the physiology is concrete. Resting core temperature is 0.3 to 0.6°C higher in the luteal phase due to progesterone. In hot conditions, that gap matters: your body starts each effort slightly more heat-loaded. Consider pre-cooling, conservative early pacing, and extra hydration if you're racing in heat during late luteal.

What does cycle irregularity mean for training?

Treat it as a nutrition alarm, not a training variable. Irregular cycles are a primary clinical indicator of Relative Energy Deficiency in Sport (REDs). The 2023 IOC REDs consensus found REDs indicators in 23 to 80% of female endurance athletes. If your cycle becomes irregular, investigate your energy availability before adjusting your training phases.

#menstrual-cycle#female-athletes#hormones#hrv#iron-deficiency#hormonal-contraceptives#training-periodization#n-of-1

Find your actual pattern, not the population average

AthleteOS lets you log cycle phase alongside every session and surfaces your personal HRV and RPE correlations across months. No imposed templates. Your data, your signal.

Generate Your Free AI Plan
14-day free trial · No credit card required