Menstrual Cycle Phase and Strength: What Studies Really Show

A 2020 systematic review and meta-analysis by McNulty et al., pooling data from 78 studies on exercise performance across the menstrual cycle, reached a conclusion that surprises many coaches: the average effect of cycle phase on performance is trivial to small, and most underlying studies were methodologically too weak to detect a real effect even if one existed. This sits uneasily alongside popular claims that female athletes should lift heavier in the follicular phase and back off in the luteal phase. The truth, based on the controlled literature, is more nuanced than either the strong pro-periodization or strong null-effect narrative suggests.

This review synthesizes strength- and power-specific research from 1996 to 2021, covering maximal force production, rate of force development, jump performance, and whether phase-based resistance training periodization produces measurably better outcomes than standard programming.

Cycle Phases and the Underlying Hormonal Fluctuations

In an idealized 28-day eumenorrheic cycle, estrogen and progesterone follow distinct trajectories that researchers use to define discrete phases for study purposes. The early follicular phase (roughly days 1 to 5, coinciding with menstruation) has low concentrations of both hormones. Estrogen then rises steadily, reaching a sharp preovulatory peak around days 12 to 14 that is 3 to 6 times higher than early-follicular baseline, while progesterone stays low. Following ovulation, the early-to-mid luteal phase brings a secondary estrogen rise alongside a much larger progesterone rise, peaking around days 20 to 22. In the late luteal (premenstrual) phase, both hormones decline sharply before menstruation resumes.

Because estrogen affects neuromuscular excitability, motor unit recruitment, and collagen synthesis in animal and cell models, researchers have long hypothesized these swings should translate into measurable, cycle-synchronized changes in strength and power in humans. Testing that hypothesis rigorously has proven far harder than stating it.

Why This Research Area Is So Methodologically Difficult

Elliott-Sale et al. (2021) published a widely cited methodological consensus statement after reviewing the female-athlete exercise science literature and found that fewer than 15% of published menstrual cycle studies verified cycle phase using a serum hormone assay or a urinary LH surge test. The remainder relied on calendar counting, which has well-documented error, since cycle length varies by 2 to 9 days even in regularly cycling women, and anovulatory cycles (no luteal progesterone rise at all) are common even in athletes who report normal periods.

This matters for interpreting the strength and power literature: a study labeling a session as luteal-phase by calendar day alone may in fact be testing a participant who did not ovulate that cycle. Sample sizes compound the problem — many foundational studies, including Sarwar et al. (1996), enrolled fewer than 10 participants, sufficient to detect only very large effects and prone to unstable estimates.

Blagrove et al. (2020), in a meta-analysis focused on strength-related measures, noted substantial heterogeneity between studies in phase definition, verification method, exercise mode, and outcome measure — heterogeneity high enough that pooling the data required caution. This is the backdrop against which every finding below should be read.

What the Evidence Shows for Maximal Strength

Blagrove et al. (2020) pooled strength-specific outcomes (isometric and dynamic maximal voluntary contraction, 1RM measures) from hormone-verified and calendar-based studies and found a pooled standardized mean difference of approximately 0.06 between the lowest-hormone (early follicular) and highest-hormone phases — a trivial effect with a confidence interval crossing zero. Their conclusion: at the group level, menstrual cycle phase does not produce a meaningful, consistent change in maximal strength in eumenorrheic women.

This population-level null finding coexists with individual studies reporting real within-subject differences. The classic Sarwar et al. (1996) study, though limited to seven participants, found isometric quadriceps maximal voluntary contraction roughly 11% higher in the luteal phase than the early follicular phase, alongside a slower voluntary relaxation rate, attributed to progesterone's effect on calcium handling in skeletal muscle. Sung et al. (2014), studying the acute response to a single resistance session, found no significant within-session strength difference by phase, but did find greater exercise-induced muscle damage markers (creatine kinase, delayed-onset soreness) when the session fell in the low-progesterone follicular phase versus the luteal phase.

Taken together: cycle phase is very unlikely to produce a strength change large enough to warrant a blanket programming rule for all female athletes, but individual women may show real, reproducible fluctuations that a population average washes out. A trivial population effect does not prove the individual effect is zero for every athlete — it proves no single across-the-board rule is supported by the aggregate data.

Power Output, RFD, and Jump Performance Across the Cycle

Power and rate of force development (RFD) research tells a similarly mixed story, with one notable exception: the longest, best-designed periodization trial in this area found a real training-outcome difference, not just an acute performance blip. Wikström-Frisén et al. (2017) randomized recreationally resistance-trained women into a 4-month structured intervention comparing phase-based periodization (heavier loads, lower volume in the follicular phase; higher volume, lower relative load in the luteal phase) against a non-periodized comparison group with matched total volume distributed evenly across the cycle. The periodized group produced meaningfully larger gains in leg-press power and squat 1RM over the full intervention — a training-adaptation effect built up across many cycles, not a single-session difference.

This matters: a null finding for acute, single-day strength testing (as in most studies pooled by Blagrove et al.) does not rule out a real cumulative training effect when load and volume are deliberately structured around the cycle across months, since fatigue management and recovery capacity — not just raw force output on a given day — determine long-term adaptation.

Julian et al. (2017), studying elite female soccer players with hormone-verified phases across a competitive season, found no significant cycle-phase effect on sprint speed, countermovement jump height, or repeated-sprint ability — plausibly because high training status and controlled testing conditions reduce cycle-related fluctuations compared to less-trained populations. Self-reported readiness and perceived exertion, however, consistently show more cycle-related variation than objective power output does, with the late luteal (premenstrual) window most often flagged as highest perceived fatigue regardless of measured neuromuscular output.

Practical Implications for Periodization

A defensible practical framework has three parts. First, do not assume a universal rule that all female athletes should train maximal strength in the follicular phase and reduce load in the luteal phase — the population-level data does not support this as default policy, and imposing it uniformly may disrupt training for athletes with no personal pattern at all. Second, treat the Wikström-Frisén et al. (2017) result as evidence that a structured, individualized approach can work over a multi-month block when built from an athlete's own observed pattern rather than a generic template, since the benefit came from a consistent, planned load-volume structure rather than reacting day-to-day to symptoms. Third, weight subjective readiness (sleep, perceived exertion, self-reported symptoms) alongside objective output, since subjective fatigue consistently varies more across the cycle than measured force or power does — perceived difficulty on a heavy day is real even when bar speed or jump height has not changed.

In practice, a starting protocol looks like four steps, repeated across at least two to three complete cycles before anyone touches the actual training program:

If the same phase shows a repeated, consistent shift in objective output across two or three cycles, that pattern is worth building a program around. If the data is noisy with no repeating shape, standard progressive programming without cycle-based adjustment remains the better default — chasing a signal that is not actually there tends to create more disruption than it resolves.

Individual Variability and the Case for Objective Monitoring

The apparent contradiction between trivial population-level effects and strong individual anecdotes is not actually a contradiction once heterogeneity of response is accounted for. Progesterone and estrogen receptor density varies between individuals, ovulatory status varies substantially even among women who report normal periods, and self-reported symptom severity does not correlate tightly with measured hormone concentrations in most studies. A meta-analysis correctly reports that the average woman shows little change while still being consistent with a meaningful subset of individual women showing real, reproducible fluctuations in either direction.

The practical consequence: phase-based periodization decisions are better made from an individual athlete's own longitudinal data than from population averages in either direction — neither assuming a strong cycle effect exists nor assuming it categorically does not. Objective session metrics such as mean concentric velocity, jump height, and estimated 1RM trend, tracked across multiple cycles alongside verified cycle day, let a coach test the hypothesis for a specific athlete rather than apply a rule derived from studies with the limitations outlined above. Where a genuine pattern emerges, adjusting programming around it is reasonable; where it does not, standard progressive programming without cycle-based modification remains well supported.