Masking, gender and diagnosis

More boys than girls are diagnosed with autism. A meta-analysis of 54 prevalence studies found an overall ratio of about 4.2 boys to 1 girl among children who meet the criteria, with large differences between studies. The ratio was lower in high-quality studies (3.32 to 1) and in studies that screened whole populations (3.25 to 1) than in studies that counted only children already diagnosed (4.56 to 1). The authors concluded that the true ratio is closer to 3 to 1 and took this as a sign of diagnostic gender bias.¹

In a 2020 narrative review, Hull, Petrides and Mandy set out two explanations for the gap. One is a "female protective effect", something about being female that makes autism less likely. The other is that autistic girls and women go unrecognized because their autism looks different from what the criteria describe. They called the second possibility the female autism phenotype, with camouflaging as one of its parts, and concluded that the evidence so far supports a female-typical presentation, while noting that it still needs study across all genders and ages.² A large genetic study supports a protective effect: in Danish registers, siblings of autistic girls and women were more often autistic than siblings of autistic boys and men, and in US family collections mothers of autistic children carried more common genetic likelihood for autism than fathers.³ A later meta-analysis of 67 studies found small differences in both directions: autistic males scored as more affected on some clinician-rated measures, autistic females scored higher on a parent- or teacher-rated scale of social difficulty, and most other measures, including repetitive behaviors and the Wechsler IQ scales, showed no sex difference. Its authors read the results as support for a clinical bias toward males and for taking a female phenotype, "potentially involving camouflaging", into account in diagnosis.⁴ An umbrella review in 2026 analyzed 34 studies, themselves systematic reviews, meta-analyses and scoping reviews. It found no conclusive evidence of sex or gender differences in social communication, friendships, play or motor stereotypies, but there was evidence that autistic females showed fewer restricted and repetitive behaviors (the earlier 2025 meta-analysis of primary studies had found no significant difference there) and were more capable at camouflaging, although autistic males camouflaged too. Its authors concluded that the sex and gender differences reported among autistic people resemble those in the general population and are not specific to autism, and that age and general intellectual ability appear to interact with some of them.⁴² In one early interview study, 14 women diagnosed in late adolescence or adulthood described "pretending to be normal" and how their gender led professionals to miss their autism.⁵

Critics question parts of this picture. Fombonne (2020) challenged the idea of camouflage itself, including how it is measured, whether it explains late diagnosis, and whether it belongs to a female phenotype.^6,7 Lai and colleagues replied that camouflaging "is not necessarily a feature of autism in female individuals", although sex and gender still shape how autism shows.⁷ Pearson and Rose argue that tying masking to a "female autism phenotype" imposes gender stereotypes and could make diagnosis harder for nonbinary people, and for men and women who do not fit the criteria.⁸ Rippon points out that the evidence base itself is skewed. In a commentary she reports that of more than 120 brain-imaging studies of social processing in autism, nearly 70% included only males or just one or two females, and that diagnostic procedures were developed mainly on male-only groups.⁹

In the adults and adolescents studied so far, mostly without intellectual disability, autistic women report more camouflaging than autistic men on average. A meta-analysis of 11 studies found this on the CAT-Q, the most widely used questionnaire (a small effect); a separate analysis in the same review, of only four studies with signs of publication bias, found females higher on discrepancy measures too.¹¹ A second meta-analysis, of 10 studies using the same questionnaire, mostly in adults, found autistic females reported somewhat more camouflaging overall; in the six studies that reported subscales, they scored higher on compensation and masking but not assimilation.⁴ A German study of 164 autistic and 482 non-autistic adolescents found the questionnaire worked the same way for both sexes, and girls scored higher only in the autistic group.¹² In one large online sample of adults, non-autistic women and men did not differ either.¹³ How strongly autistic traits go with camouflaging is similar in both sexes.¹⁴ A critical review of 389 studies notes that the samples overrepresent women diagnosed in adulthood and underrepresent autistic people with intellectual or language disabilities or higher support needs.¹⁵ Clinical diagnosis data show a similar drift with age. In a Scottish clinical sample of 408 people assessed for autism, ADHD or intellectual disability across 30 teams, the male-to-female ratio among those diagnosed autistic was 2.21 to 1 under age 10, but there were more females than males in adolescence (0.79) and adulthood (0.94). Autistic female children and adolescents were referred later than males (median 10.5 versus 7.5 years) and diagnosed later (11.9 versus 9.3 years), while there was no sex difference in the age of referral or diagnosis among autistic adults. The authors read the pattern as girls being missed while young.⁴³

The CAT-Q's developers reported that its structure held across genders and across autistic and non-autistic adults, in an online sample of 354 autistic and 478 non-autistic people.²³ An independent German study of adolescents found the same for sex and age group.¹² Other checks are less tidy. A critical review of 389 studies judged the support for the CAT-Q's validity mixed, because it overlaps with other traits such as social anxiety and is not specific to autism, and it also questioned how discrepancy measures are built.¹⁵ In a Swedish sample aged 10 to 83, the original three-part structure did not fit well and validity was weaker in children under 15, although autistic females still scored higher than autistic males.²⁴ In 308 autistic adults, most items stood apart from social anxiety and fear of being judged, but the assimilation items also loaded with autistic social-trait items.²⁵ In a representative sample of 972 US adults, the CAT-Q showed nearly the same structure as in autism-focused samples, which suggests it also captures something many non-autistic people do.²⁶

Self-report captures how much people try to camouflage, not how well it works. In one study of young adults, self-reported camouflaging correlated with only one of two discrepancy measures.¹⁹ The wider autism questionnaires raise a related worry. In interviews and surveys with 22 autistic women and gender-diverse people assigned female at birth, researchers found a shared view that common screening questionnaires measure "only one way to be autistic".²⁷ A systematic review of whether adult autism self-report tools work equally across genders has been registered; its results had not yet appeared in any study found as of October 2026.²⁸

Across five datasets totaling 641,860 people, transgender and gender-diverse people were more often diagnosed autistic than cisgender people, and on average scored higher on self-reported autistic traits, whether or not they were autistic.²⁹ How far autistic people in this group mask, what they say about it, and what the research says about gender-affirming care are covered in Gender diversity and trans autistic people.

Whether girls are diagnosed later than boys varies between studies: a systematic review of 56 studies from 40 countries found that studies reporting on factors such as gender often disagreed or were inconclusive about their effect on age at diagnosis.³³ In one Ontario clinical sample of 365 children (20% girls), parents first became concerned at similar ages for girls and boys, but girls were diagnosed later, and higher verbal ability went with later diagnosis more strongly in girls.³⁴ A systematic review of barriers for girls and young women under 21 found that parents saw compensatory behaviors among them, along with other people's perceptions and clinician bias.³⁵

In two samples of cisgender autistic adults, women were diagnosed later than men, though in the larger online sample of 812 adults only after adjusting for age and autistic traits, and in the small longitudinal sample of 59 only before adjusting. Camouflaging went with later diagnosis more strongly in women than in men only in the small sample, and within each sex it did not predict age at diagnosis; the authors call the findings tentative.³⁶ Among 502 autistic adults, those diagnosed as adults reported more assimilation and compensation than those diagnosed in childhood.¹⁷ In a study of 80 adults, how much people meant to camouflage did not predict strangers' first impressions of them, but better first impressions went with a later diagnosis.³⁷ Another study, of 117 adults (58 autistic, mostly intellectually able women), found the number of compensation strategies people reported was unrelated to age at diagnosis or to sex.²⁰ In 174 young adults from a UK twin study, those with high autistic traits but no diagnosis had larger gaps between their traits and how autistic they appeared than diagnosed adults did, though both groups reported similar camouflaging. On at least one measure the difference held only for men. The authors offer one interpretation, that effective camouflaging may lower the chance of diagnosis.¹⁹

Because these studies measured camouflaging after diagnosis, they cannot show that camouflaging delayed it. In the study of 733 children and adolescents, the gap between parents' and clinicians' ratings of autistic traits, a sign that a clinic visit may show less than daily life, was larger in adolescents than in younger children.²¹

The DSM-5 allows for masking in its criterion C: "Symptoms must be present in the early developmental period (but may not become fully manifest until social demands exceed limited capacities, or may be masked by learned strategies in later life)."³⁸ Livingston, Shah and Happé note that no guidelines exist for detecting these strategies, or for whether they should be encouraged.³⁹

The main observational assessment, the ADOS, may also read girls and boys differently, for reasons that are not all masking. In 4,550 toddlers from studies of younger siblings of autistic children, girls with an autistic older sibling were rated as less affected than boys on some items (eye contact, response to joint attention and the quality of social approaches) at the same underlying level of social communication difficulty.⁴⁰ At 20 to 40 months these children are too young for learned strategies; the authors relate the gap to sex differences in social communication in the general population, which diagnostic thresholds do not account for.⁴⁰ In 813 school-age children, six items scored differently by sex. On four social communication items girls were rated as showing fewer autistic features, and on two items about repetitive behaviors and interests they were rated as showing more. The overall effect was small, about 0.41 points, and unlikely to change who met the cutoff. The study joins three earlier reports, with two items recurring, and its authors only speculate that masking or clinician bias lies behind the pattern.⁴¹ A 2026 preprint pooled item-level ADOS data from 19,569 people aged 12 months to 75 years, across four data sources and every module, which its authors describe as the first analysis of sex bias in the ADOS across all ages and ability levels. Eleven of 112 items scored differently by sex after correction; there were none in the Toddler Module, Module 1 Low Language or Module 2, and the most in Module 4 (7 of 14 items). Adult females tended to be rated as showing fewer social communication difficulties and more restricted and repetitive behaviors, and every effect was small except one (stereotyped language, Module 1 High Language). The authors concluded that the ADOS is largely robust for younger children and those with less language, but shows bias for verbally fluent adolescents and adults. The preprint has not been peer reviewed.⁴⁴

Devon Price, who is transgender and autistic, notes that researchers, therapists and some autistic self-advocates talk about "female Autism", and calls the usual trait lists "very broad, very gendered". Price argues the label misleads because it locates masking in a person's sex rather than in social expectations: "Masking is a social experience, not a biological one." Price says their own experience fits neither the "female" nor the "male" autism story, and quotes Bobbi, a nonbinary interviewee: "I wasn't raised or 'socialized' as an Autistic girl. I was raised as a weird kid, and a gender failure." Price also describes autigender, a term some autistic trans people use for a gender identity they see as bound up with being autistic.^c1

Eric Garcia writes that the concept of being autistic "is biased toward males", and quotes the autistic scholar M. Remi Yergeau on assessment: "What clinicians see is the masking."^c2 Sol Smith recalls a doctor who doubted Smith was autistic because "you make pretty good eye contact", although Smith had "a long list of eye contact work-arounds".^c3 Megan Griffith describes how people who mask heavily are told after diagnosis that they seem too "normal", which feels like a denial of their past struggles.^c4 Megan Anna Neff writes that for some autistic people, fitting cisgender and heterosexual norms has been part of masking, so working out their gender and sexuality can come later and be harder.^c5

Amy Pearson, an autistic psychologist, and Kieran Rose, an autistic trainer and researcher, argue that the link between gender and masking is "unlikely to be driven by a 'female phenotype'". In their reading it comes from living where being autistic meets being female, or meets a particular upbringing and culture. Seen that way, they write, missed diagnoses can be understood "without unintentionally blaming women" for them; and since most masking research has studied white women, treating masking as a female trait risks "another white-centric narrative".^c6 Carly Jones, an autistic writer on safeguarding girls, asks whether "rigid stereotypes of both gender and Autism" keep a larger group from diagnosis, and describes a gay man whose doctor turned him away because he did not "look" autistic, "as can so often be the case for females".^c7

Amy Sequenzia, a nonspeaking autistic writer, puts girls' later diagnosis down to "society's expectations of how girls should behave, be, or react", and recalls being called an "atypical" autistic because she liked to cuddle and could make eye contact with some people.^c8