Sex of a Skull: How to Tell Male From Female

Forensic anthropologists distinguish male from female skulls by examining a handful of structural features shaped largely by hormones during puberty: the brow ridge, the mastoid process behind the ear, the jawline, and the overall size and ruggedness of the cranium. When an experienced analyst has a complete adult skull, accuracy typically falls in the range of roughly 80 to 90 percent, though that number shifts depending on the population the skull comes from, how much of the skull is available, and whether traditional visual methods or newer computational techniques are being used. The skull is the second most reliable skeletal element for sex estimation after the pelvis, and in many forensic and archaeological scenarios it is the only element available.

The Five Classic Traits

The most widely used visual method asks an observer to score five cranial landmarks on a scale from clearly female to clearly male. Each trait reflects differences in bone robustness, muscle attachment size, or overall cranial proportions. In general terms, male skulls tend to be larger, heavier, and have more pronounced surface features, while female skulls are smoother, lighter, and more rounded.

• Nuchal crest: The ridge of bone at the back of the skull where the neck muscles attach. In males it is usually more prominent and rougher; in females it tends to be smooth and barely palpable.
• Mastoid process: The bony bump behind the ear. Males typically have a large, bulky mastoid, while in females it is smaller and more pointed. This is one of the most consistent individual traits for sex estimation.
• Supraorbital margin: The upper rim of the eye socket. A sharp, thin edge suggests female; a thick, rounded, blunt edge suggests male.
• Glabella: The flat or protruding area between the brow ridges, just above the nose. Males tend to have a prominent, projecting glabella, while in females it is flat or only slightly raised.
• Mental eminence: The chin. Males generally have a broader, more square chin with a prominent central projection, while females tend toward a narrower, more pointed chin.

This scoring system, formalized by Buikstra and Ubelaker and later refined by Walker, is the most common nonmetric method used by physical and forensic anthropologists. Studies of inter-observer reliability find good agreement on most traits, with one persistent weak spot: the mental eminence, which shows considerable variation in how different observers score it.¹ That same pattern appears across multiple independent assessments, making the chin the least dependable of the five landmarks for consistent scoring.²

What the Jawbone Reveals

The mandible carries its own set of sex-related differences, independent of the cranium above it. Males tend to have a wider jaw measured at the gonial angles (the corners where the jaw turns upward), taller ramus branches on each side, and more prominent condylar and coronoid processes at the top of the ramus. Females generally show a larger mandibular angle, a wider ramus relative to its height, and a more gracile overall body.³ Simple metric measurements like mandibular length, bigonial breadth, and bicondylar breadth all show statistically meaningful differences between sexes.⁴

These shape differences are not simply a scaled-down version of the cranial ones. Research on mandibular development shows that sexual dimorphism concentrates at the ramus and the chin region early in growth, fades in childhood, and reasserts itself at puberty, and this pattern is not driven by dental development.⁵ Still, the mandible used alone is less reliable than either the full cranium or the pelvis. A study from the Low Countries found that mandible-based sex assessment agreed with pelvis-based assessment only about 70 percent of the time, and among individuals with female pelves, over half of the associated mandibles were scored as male.⁶ The full cranium fared much better, agreeing with the pelvis in about 96 percent of cases in the same study.

How Accurate Is the Skull Compared to the Pelvis

The pelvis is the gold standard for skeletal sex estimation because its shape is directly shaped by the functional demands of childbirth. Pelvic features like the subpubic angle and the sciatic notch routinely achieve classification accuracy above 90 percent. The skull, by contrast, reflects sex differences that are more about overall body size, muscle mass, and hormonal influence on bone remodeling, so the signal is noisier. In a large Balkans forensic sample, the pelvis was correctly sexed in about 93 percent of cases, while the skull alone was correct in roughly 71 percent, with an additional 15 percent of skulls judged ambiguous.⁷

That 71 percent figure, though, is context-dependent. It represents a setting where many remains were fragmentary and came from a conflict zone. Under better conditions and with population-specific reference data, skull-based methods perform considerably higher. A discriminant-function analysis of Japanese skulls achieved about 86 percent accuracy.⁸ A Brazilian study using a different set of craniometric measurements reached about 72 percent with just two measurements, but the authors noted that adding more variables would push accuracy higher.⁹ The takeaway is that skull-based methods work, but they are not as reliable as the pelvis, and the exact accuracy you get depends heavily on which population’s standards you apply.

Why Population Background Matters So Much

One of the biggest pitfalls in skull-based sex estimation is using standards developed on one population to assess individuals from another. The degree to which male and female skulls differ varies across populations, and so do the specific features that differ most. Geometric morphometric studies confirm that both sex and population ancestry produce significant differences in cranial shape, and these effects overlap in ways that can easily confuse a method tuned to the wrong group.¹⁰

A clear illustration of this came from applying Walker’s method, developed primarily on North American samples, to a group of South Asian individuals whose chromosomal sex was confirmed through protein analysis. Overall accuracy was only about 63 percent, and for males specifically it dropped to roughly 36 percent, while all females were classified correctly.¹¹ The method was effectively reading smaller-framed South Asian males as female because their cranial robustness fell below the thresholds calibrated for a different population. Population affinity shapes how sexually dimorphic traits manifest in bone, influenced by body size, diet, activity patterns, and genetic background, which means ancestry estimation is a necessary step before sex estimation in forensic casework.¹²

Even specific anatomical landmarks can vary in their usefulness across groups. A study examining the krotaphion-sphenion distance in the temple region found statistically significant differences across German, Euro-American, African-American, and Rwandan skulls for both sex and ancestry, illustrating that a measurement useful in one group may not carry the same discriminating power in another.¹³

What Testosterone Does to the Skull

The sex differences in skull morphology are not arbitrary. They trace back to hormonal influences during development, primarily testosterone. During adolescence, rising testosterone levels drive the growth of the brow ridges, the enlargement of the mastoid process and jaw, and the overall increase in cranial robustness that distinguishes adult male skulls. Research on adolescent faces has shown that the features most reliably perceived as male are associated with plasma testosterone levels, independent of age.¹⁴

Animal experiments strengthen this connection. When testosterone was suppressed in developing animals, craniofacial growth was measurably altered across the cranial bones, maxilla, and mandible. Condylar cartilage development slowed, bone mineralization was delayed, collagen fiber synthesis dropped, and vascularization in cranial sutures decreased.¹⁵ These findings help explain why the sex differences forensic anthropologists rely on are products of puberty: they emerge when hormonal exposure diverges between males and females, and they are not present in meaningful form before that divergence occurs.

Masticatory forces also play a shaping role. The muscles you use to chew exert mechanical loads on the skull throughout life, influencing the transversal and vertical dimensions of the face. Stronger chewing muscles are associated with broader bone bases, more developed gonial angles, and more prominent condylar and coronoid processes on the mandible.¹⁶ Because males on average have greater muscle mass including masticatory muscles, these biomechanical forces compound the hormonal signal, making male jaws and cheekbones not just hormonally but mechanically different from female ones.¹⁷

Children’s Skulls Are a Different Problem Entirely

Because so much of cranial sexual dimorphism depends on pubertal hormones, estimating sex from the skulls of children and adolescents is far less reliable. Before puberty, the skeletal differences between boys and girls are subtle and heavily overlapping. Classification accuracies for subadult remains generally hover between 68 and 72 percent, which is barely better than a coin flip.¹⁸ Many researchers treat sex estimation from juvenile skeletal remains as inherently unreliable, and forensic reports on children’s remains often leave sex as “indeterminate” rather than risk a misclassification.

How Aging Blurs the Differences

At the other end of the age spectrum, aging gradually erodes some of the cranial features that distinguish males from females. After about age 60, both sexes undergo widening of the neurocranium and retrusion of the face, including the forehead. When skull shape data are adjusted for overall size, the sexual dimorphism of the frontal, occipital, and zygomatic regions tends to diminish in elderly individuals. This age-related convergence degrades the accuracy of sex classification by roughly two to three percentage points compared to younger adults.¹⁹

Tooth loss compounds this problem. When teeth are lost during life, the alveolar bone that supported them gradually resorbs, changing the shape of the lower face, the upper jaw, and the palate. These changes can shift both metric measurements and qualitative trait assessments used for sex and ancestry estimation.²⁰ An elderly skull with extensive tooth loss can look noticeably different from what the same individual’s skull looked like at age 30, making it harder to score and easier to misclassify.

Working With Fragments

In real forensic and archaeological cases, you rarely get a pristine, complete skull. Remains are often fragmented, weathered, or incomplete, and the analyst has to work with whatever pieces are available. Different skull regions carry different amounts of sex information. A study of Brazilian skulls tested four geometric triangle methods on different cranial areas and found wide variation: the mastoid region gave about 80 percent accuracy, while the facial and occipital triangles hovered around 63 to 64 percent.²¹

The temporal bone, which houses the mastoid process and sits on the side of the skull near the ear, is among the most useful isolated elements. Its dense bone makes it relatively resistant to destruction, and it carries several sexually dimorphic features. A study on an Egyptian population developed a sex-prediction equation based on the mastoid, styloid, and zygomatic processes of the temporal bone, finding that all measured parameters were significantly larger in males.²² If you have only a fragment of the skull, the temporal region is one of the better pieces to have.

Machine Learning and 3D Scanning

Traditional visual scoring depends on the analyst’s training and experience, which introduces subjectivity. The push in recent years has been toward computational methods that can standardize the process and potentially push accuracy higher. Three-dimensional scanning, whether from CT data or surface laser scans, allows researchers to capture the full geometry of a skull and apply statistical algorithms to the resulting shape data.

A deep-learning neural network trained on images of 900 skulls reconstructed from hospital CT scans achieved 95 percent accuracy on previously unseen skulls.²³ An unsupervised machine-learning approach using 78 landmarks across the skull surface reported even higher numbers, reaching 98 percent for females and 93 percent for males.²⁴ Another method using partial least-squares dimensionality reduction on 3D skull shapes achieved 100 percent accuracy across all cranial regions examined, though such perfect results should be treated cautiously since they may reflect overfitting to the training sample.²⁵

These numbers are impressive, but they come with caveats. Most of these systems are trained on skulls from a single population or a narrow set of populations, so their performance on skulls from other backgrounds is unknown until tested. They also require expensive scanning equipment and computational infrastructure that many forensic laboratories, especially in resource-limited settings, do not have. The potential to eliminate human bias is real, though. Once trained, an algorithm applies the same criteria to every skull, which removes the observer-experience variable that plagues visual methods.

When Proteins Do What Bones Cannot

A completely different approach sidesteps morphology altogether. Bone contains proteins, and certain proteins are encoded by genes on the sex chromosomes. By extracting and analyzing peptides from a bone fragment using mass spectrometry, researchers can identify whether the individual had XX or XY chromosomes, providing what is effectively a molecular determination of chromosomal sex.

This technique has proven especially valuable in cases where morphological methods struggle. In a highly commingled ossuary in Spain, a peptidomics approach showed 100 percent agreement with osteological sex estimates when both skull and pelvis traits were available.²⁶ In a separate study, peptide-based analysis both increased the number of individuals who could be assigned a sex and revealed six cases where osteological and molecular results disagreed.²⁷ Those disagreements are revealing: they show that some individuals whose bones look male had XX chromosomes, or vice versa, which raises questions about how well morphological methods capture the full spectrum of human variation.

Protein-based methods are especially useful for juvenile remains, fragmentary bones, and populations where morphological standards have not been developed. They are destructive, though, requiring a small sample of bone to be consumed in the analysis, and they need specialized laboratory equipment. For archaeological remains, there is also the question of protein preservation over centuries or millennia. Still, the technique represents a meaningful shift in how sex estimation can be validated and, in difficult cases, overridden.

A Pathological Clue Hidden Inside the Skull

One quirk of skeletal biology occasionally offers an unexpected assist. Hyperostosis frontalis interna, or HFI, is a condition where bone thickens on the inner surface of the frontal bone. It is not a disease in the usual sense; most people with it never know. But it is strongly sex- and age-linked. HFI is significantly more prevalent in females than males, and in both sexes it becomes more common with age.²⁸ The severe form of HFI is almost four times more common in women than men.²⁹

The association appears driven by hormonal factors, particularly androgens and dysregulation of insulin-like growth factor, with increased prevalence among older females over time.³⁰ For a forensic examiner, finding severe HFI in a skull that also shows markers of advanced age tips the probability strongly toward a female decedent. It is not definitive on its own, but in an ambiguous case it provides a useful additional data point.

From Skull to Face

Sex estimation from the skull also feeds into forensic facial reconstruction, the process of building an approximate face on an unidentified skull to aid in recognition. Knowing whether a skull is male or female changes the soft tissue depth assumptions the reconstructor applies and the overall facial proportions they model. Recent 3D vector-based research on the relationship between skull landmarks and corresponding skin-surface points found statistically significant sex differences at most measured locations, though only a few differences exceeded the measurement error threshold in practice.³¹ Getting the sex wrong at the skull stage cascades into a facial approximation that may look like the wrong person entirely, which is why forensic anthropologists treat sex estimation as a foundational step before any reconstruction work begins.

In court settings, the methods used for sex estimation must meet legal standards for the admissibility of scientific evidence.³² That means the analyst has to be able to articulate the error rate, the population the method was validated on, and whether it has been peer reviewed. The push toward computational and molecular methods is partly about improving accuracy, but it is also about producing results that stand up to legal scrutiny in ways that subjective visual scoring sometimes struggles with.