Few claims from evolutionary psychology have escaped the academy as thoroughly as this one: that women's taste in men swings with their menstrual cycle, tilting toward square jaws and heavy brows when they are most fertile and toward softer, kinder faces the rest of the month. It appears in undergraduate lecture slides, dating advice columns and podcast monologues. It is also, in its strong form, one of the more instructive casualties of psychology's replication crisis.

The more interesting story is what remains standing after the wreckage — a narrow, weak hormonal signal, a surprisingly modest average preference for masculine features, and a beard literature that turns out not to run parallel to the masculinity literature at all.

The Hypothesis That Launched A Thousand Studies

The idea has a formal name: the ovulatory shift hypothesis. It proposes that women experience evolutionarily adaptive changes in mating-related thoughts and preferences across the ovulatory cycle. The specific prediction is that preferences for masculine physical and behavioural traits should be strongest during the peri-ovulatory window, when conception is possible, and weaker at other points.

The reasoning behind it was a dual mating strategy. Near ovulation, women were thought to favour men displaying cues of "good genes" — facial masculinity, dominance, symmetry. Outside that window, the argument went, they would favour "good father" traits signalling resources and willingness to invest. Facial masculinity earned its place as a candidate cue through the immunocompetence handicap idea: because testosterone suppresses immune function, only men with genuinely robust constitutions could afford to develop heavily masculinised features without succumbing to infection. A strong jaw, in this telling, was an honest advertisement.

The foundational empirical work came from Ian Penton-Voak, David Perrett and colleagues around 1999, followed by Anthony Little and others in 2002, using computer-morphed faces in which masculinity could be dialled up or down along a continuous scale. Those studies reported that women in the fertile phase preferred relatively more masculinised faces, and that the effect was clearest when women were evaluating men as short-term rather than long-term partners.

It was an elegant result. It was mechanistic, it made falsifiable predictions, and it seemed to explain something people already believed they observed. It was cited relentlessly.

The Cracks Appear

Trouble arrived early and from several directions at once.

In 2009, a study published in PLOS One departed from the standard method by using photographs of real individual men rather than computer-generated composites. Twenty-five normally cycling women rated faces and bodies at high- and low-fertility days. The researchers found no evidence of any cyclic shift in preferences for masculinity or symmetry. Their conclusion was pointed: the subtle shifts detected with artificial stimuli might simply not survive contact with real faces, and therefore might not influence actual mate choice.

A year later, Christine Harris published a considerably larger test — 853 adults evaluating a substantially bigger pool of target faces, with a more thorough assessment of ovulatory status than earlier work had used. The results showed no greater preference for masculine faces when fertilisation was likely. Harris also argued that the theory rested on questionable assumptions about ancestral human mating systems.

The response was swift. Lisa DeBruine, Benedict Jones, Martie Haselton, Penton-Voak, Perrett and David Frederick published a rebuttal arguing that Harris had misrepresented the size of the existing literature, that many more studies existed than she acknowledged, including one substantially larger than hers, and that considered as a whole the evidence for cyclic shifts remained compelling despite one failed replication.

That standoff hardened into open warfare in 2014, when two meta-analyses of the same broad literature reached opposing conclusions about whether the evidence was robust enough to support cycle-linked preference shifts at all. One camp read the aggregate data as confirming the effect. The other read it as an artefact of small samples and publication bias.

Underneath the dispute lay a set of methodological problems that both sides largely acknowledged. Many studies estimated cycle phase from self-reported menstrual dates rather than confirming ovulation hormonally. Samples were often tiny. Designs were frequently between-subjects rather than within-subjects, comparing different women at different phases instead of the same woman across her own cycle. And the stimuli were usually synthetic.

The Decisive Test

The most consequential study arrived in 2018 in Psychological Science, and its significance was partly sociological. The lead authors — Benedict Jones and Lisa DeBruine — had spent years publishing supportive work in this area. This was not a takedown by hostile outsiders.

Their team ran the largest longitudinal study of the hormonal correlates of masculinity preference ever conducted: 584 heterosexual women, tested in repeated weekly sessions, with salivary steroid hormone levels measured directly rather than inferred from calendar dates. The analyses showed no compelling evidence that preferences for facial masculinity tracked changes in women's hormone levels. Neither within-subjects nor between-subjects comparisons showed any evidence that oral contraceptive use lowered masculinity preferences — a related claim that had circulated widely on its own.

Jones summarised it plainly at the time: with much larger samples and direct measures of hormonal status, they could not replicate the effect.

Subsequent work has largely followed suit. A 2023 eye-tracking study collected saliva from 81 women at three points across the cycle and measured both stated preferences and where the eye actually lingered. It found no evidence that the estradiol-to-progesterone ratio predicted preferences for facial masculinity, and no evidence that mate choice shifted across the cycle — though hormones did relate to visual attention to men generally.

What Actually Survives

The picture is mostly null, but not entirely, and the residue is worth stating precisely.

Some well-powered work still detects a hormonal signal when hormones are measured directly, even where categorical cycle phase predicts nothing. One such project combined a very large self-report sample of 2,161 women with a smaller within-subjects study in which ovulation was confirmed by luteinising hormone testing, plus salivary estradiol and progesterone measured in 36 participants. Preferences did not vary with self-reported cycle phase, and they did not vary with LH-confirmed fertility. But within individual women, rising estradiol predicted somewhat stronger preferences for more masculine faces, while higher progesterone predicted a shift toward more feminine faces.

So the defensible remaining claim is not that women prefer masculine men when they are ovulating. It is that specific circulating hormones may exert a small, inconsistent influence on preference that is difficult to detect and easy to lose. That is a considerably humbler proposition than the one in circulation.

One clarification is worth making, because the popular version routinely garbles it: the predicted shift is toward the fertile mid-cycle window around ovulation. Menstruation itself is the low-fertility phase. Anyone framing the question as "does she prefer masculine men when she has her period" has the prediction backwards.

The Preference Is Smaller Than Men Think

Strip out the cycle question and a separate surprise emerges. The baseline preference for masculinity is real but modest, and there is no reliable figure describing what proportion of women hold it — studies report average positions on a continuum rather than headcounts on either side.

The closest quantitative anchor comes from a 2025 study in which participants moved a slider to adjust male facial masculinity. Women, on average, settled on faces about 33 per cent more masculine than the least masculine version available, which the authors described as only a moderate preference. Men, asked to predict what women would choose, landed near 77 per cent — a dramatic overestimate. Men also assumed women's preferences would swing hard toward masculinity for short-term partners; women's actual settings stayed relatively stable across relationship contexts.

The direction of the average preference is itself method-dependent. Some large studies find that unmanipulated, average male faces are rated most attractive, with both heavily masculinised and heavily feminised versions rated lower. Where a masculinity advantage does appear, it is small: in one sample of 919 women, very masculine faces beat the least masculine ones with effect sizes around 0.29 for short-term and 0.41 for long-term attractiveness. The 2018 Jones study did find that women generally preferred masculinised over feminised faces, and that this was stronger for short-term than long-term judgments.

The Beard Is Not The Jawline

It is tempting to fold facial hair into the same story — beards look masculine, so surely beard preference and masculinity preference are the same phenomenon. They are not, and the divergence is robust enough that Barnaby Dixson's team titled a paper around it.

Beards do increase perceived masculinity, and roughly linearly: more facial hair, more masculine the face reads. But attractiveness does not follow that line. Women's attractiveness ratings peak at intermediate levels of facial hair. In the largest study of its kind, involving 8,520 women rating faces morphed across four levels of beardedness and five levels of facial masculinity, stubble was judged most attractive overall, while full beards fared better specifically for long-term relationships. Faces at the masculine and feminine extremes were least attractive regardless of context.

Older correlational work pointed the same way from a different angle, finding that preference for less masculine facial shape and preference for light stubble were positively associated — the opposite of what a single underlying masculinity taste would predict.

The Masculinity Paradox

Dixson's interpretation is that the two traits carry different signals. Masculine bone structure has been linked to genetic-quality cues such as health, but also to lower paternal investment, and it tends to be favoured in short-term evaluations. Beardedness maps more onto apparent age, social dominance and intrasexual formidability than onto underlying health, and it tends to be favoured more strongly for long-term relationships.

The two also interact rather than add. In the 8,520-woman study, stubble and beards dampened the polarising effect of extreme facial shape: heavy facial hair softened the penalty attached to both strongly masculinised and strongly feminised faces, plausibly by masking the underlying morphology. Facial hair enhanced long-term attractiveness and not short-term attractiveness. Rather than amplifying a shared preference, beards appear to moderate the masculinity signal.

Context Beats Chemistry

Put the strands together and the variable doing the most work is not endocrine at all. Across these literatures, the split between short-term and long-term evaluation reliably shapes what women rate as attractive — masculine morphology gaining ground in short-term judgments, beards in long-term ones — while cycle phase, measured properly, produces little or nothing.

The usual caveats deserve emphasis. Nearly all of this rests on ratings of digitally manipulated photographs by young, predominantly Western participants in online studies, and a rating is not a mate choice. Effect sizes throughout are small enough that stimulus selection and sample composition move results substantially. The beard-versus-bone-structure dissociation has held up better than the cycle-shift claim, but neither should be carried far beyond the laboratory.

What is safe to say is this. The strong, popular version — that a woman's taste in male faces swings predictably with her monthly cycle — is not supported by the best-powered, hormonally confirmed research, and several of the researchers who built that case have said so themselves. A weak hormonal influence may persist at the margins. But if you want to predict how a man's face will be judged, the more useful question is not where she is in her cycle. It is what she is judging him for.