Can an autism assessment miss you if you mask?

What masking is, and how researchers measure it

Camouflaging is the term the research literature uses for the strategies some autistic people employ to cope with the everyday social world, covering their autistic differences and difficulties in the process (Cook et al., 2021).

That systematic review pulled together 29 studies and found the field measures camouflaging two ways. One approach compares how autistic someone appears from the outside with how autistic they are on an internal measure, and treats the gap as camouflage. The other asks people directly, usually through the Camouflaging Autistic Traits Questionnaire. The authors concluded these capture two connected but distinct things: an observable behavioural presentation, and a person’s own sense of the effort they’re putting in.

That distinction matters here. A clinician can observe the surface, while the effort underneath may be available only through self-report. Both belong alongside developmental history, current functioning, and the wider clinical picture; self-reported effort is additional information, not a diagnostic measure on its own.

Three findings held across much of the literature the review covered: adults reporting more autistic traits report more camouflaging, sex and gender differences in camouflaging exist, and higher self-reported camouflaging is associated with worse mental health. The same review is direct about the limits of its own evidence base, noting that participant characterisation and representativeness were poor enough that the conclusions can’t be applied to the autistic community as a whole.

A related line of work describes compensation: autistic people who show few symptoms at the behavioural surface while continuing to report autism-related cognitive difficulties. In a study of 136 adults, some diagnosed, some self-identified, and some neither but reporting social difficulties, the undiagnosed participants used compensatory strategies that were qualitatively similar to those used by diagnosed autistic adults. The authors raised the possibility that those strategies help keep some people below the diagnostic threshold, and argued that diagnostic guidelines should account for compensation (Livingston, Shah and Happé, 2019).

What the evidence links masking to

The strongest recent synthesis is a two-part systematic review and meta-analysis published in 2025. Its first part covered 67 studies comparing autistic females and males and found that males scored as more severely affected on standard clinical measures of core symptoms and social interaction, while females showed more cognitive and behavioural difficulties. Its second part meta-analysed ten camouflaging studies and found that females used more compensation and masking strategies than males. The authors read the two together as support for a bias towards males in clinical procedures, and for taking a female autism phenotype into account during diagnosis (Cruz et al., 2025).

The instruments themselves have been examined for the same pattern. One study matched 228 school-aged young people with autism diagnoses on age and IQ and compared their scores on the ADOS and the ADI-R. Overall, females and males were rated similarly on both. But females with higher IQs were less likely to meet criteria on the ADI-R, and females were rated as more affected on parent-reported autistic traits and adaptive skills than their diagnostic scores suggested. The authors concluded that some autistic females may be missed by current diagnostic procedures (Ratto et al., 2018). That sample was children and adolescents, and the effect appeared on the developmental-history interview, not across every measure.

There’s also evidence connecting how someone comes across to when they get diagnosed. In a study of 80 autistic and non-autistic adults, ten-second video clips were rated by 127 non-autistic peers who knew nothing about diagnostic status. Autistic participants were rated more poorly overall, and better first impressions were associated with a later age of autism diagnosis (Belcher et al., 2022).

Two further sources describe the same pattern from different angles. A systematic review of barriers to diagnosis for girls and young women under 21 identified compensatory behaviours, other people’s perceptions, and clinician bias among the barriers reported (Lockwood Estrin et al., 2021). And a qualitative study of 14 women diagnosed in late adolescence or adulthood collected detailed accounts of “pretending to be normal” and of the ways their gender led professionals to miss their autism (Bargiela, Steward and Mandy, 2016).

Delay before an adult diagnosis is common in clinical samples, whatever its cause. Among 161 adults who received a first autism diagnosis at two Italian university centres, the median age at diagnosis was 23 while the first evaluation by a mental health professional had happened at a median age of 13, a gap of 11 years. Two thirds had received at least one other psychiatric diagnosis first, most often intellectual disability, psychosis, a personality disorder, or depression (Fusar-Poli et al., 2022). That study measured delay and prior diagnoses in one clinical sample. It didn’t measure camouflaging, so it can’t be read as showing that masking caused the gap.

Where the evidence stops

The claim this page holds to is that camouflaging can obscure traits and is associated with later diagnosis. The stronger claim circulating online, that high-masking adults are frequently misdiagnosed, isn’t supported by what has been published. Four limits are worth naming.

Nobody has measured a miss rate. There’s no false-negative figure for adults who camouflage. The studies above show group-level patterns in scores, impressions, and diagnostic age. None of them tracked a cohort of high-masking adults through assessment and counted how often the outcome was wrong.

The mechanism isn’t as clean as it sounds. In the first-impressions study, camouflaging intent as measured by the questionnaire did not predict how participants were rated. Better impressions predicted later diagnosis, but the deliberate effort people reported wasn’t what produced those impressions (Belcher et al., 2022). The chain from intent to appearance to a missed diagnosis has a broken link in the middle of it.

Most of the evidence is self-report from unrepresentative samples. The camouflaging questionnaire literature draws heavily on adults without intellectual disability recruited online, which is the limitation the 2021 systematic review flagged about its own field (Cook et al., 2021). Non-autistic people camouflage too: in the study that introduced gender comparisons on the questionnaire, 472 non-autistic adults reported camouflaging, with no gender difference among them, while autistic women scored higher than autistic men (Hull et al., 2020).

Group averages don’t diagnose individuals. A meta-analytic finding that autistic females are rated as less severely affected on clinical measures describes a distribution. It says nothing about whether one particular assessment reached the right conclusion.

Why a negative result isn’t automatically wrong

Masking is not an unknown that assessment ignores. DSM-5-TR requires that traits be present in the early developmental period while stating that they may be masked by learned strategies later in life, or may not become fully apparent until social demands exceed a person’s capacities (American Psychiatric Association, 2022). A clinician working from the criteria is meant to have that clause in view.

Published guidance also treats the appointment as one source among several. NICE says a comprehensive adult assessment should, where possible, involve a family member, partner, carer, or other informant, or use documentary evidence such as school reports of current and past behaviour and early development (NICE, 2021).

Structured observation carries less weight on its own than people assume. A retrospective study of 88 adults referred to a single specialist NHS adult autism and ADHD service found that, in that sample, the ADOS-2 threshold had 92% sensitivity and 57% specificity against a multidisciplinary consensus diagnosis, and the authors recommended caution when interpreting an ADOS-2 score in isolation (Adamou, Jones and Wetherhill, 2021). Those figures describe one service and one sample and shouldn’t be treated as universal, but the direction is useful: a strong result on an observation tool isn’t the whole assessment, and neither is a weak one.

Which leaves a plainer position than either extreme. Masking is a reason to check that an assessment had your developmental history, your present-day life, and the effort behind your presentation in front of it. It isn’t a reason to treat every negative outcome as a false negative. Autistic traits overlap with other explanations, an assessment can conclude that another one fits better, and that conclusion can be right.

What to do if you think masking shaped your assessment

Four things are worth doing, and none of them involves performing.

Describe the strategy and its cost, not the outcome. A clinician can observe that you held eye contact. They can’t observe that you were counting seconds to know when to break it, or that you rehearsed the opening of the conversation in the car. Effort has to be reported. Concrete examples travel better than a general statement that you mask.

Bring history, not a performance. School reports, old work reviews, letters, diaries, and an informant who knew you early all serve the part of the assessment that observation can’t reach. If nobody is available, say so when you book and ask how the service handles developmental history in that situation. Our page on what happens in an adult autism assessment covers the stages this fits into.

Say what happens afterwards. Recovery time, shutdowns, and the mental-health cost of sustained camouflaging are part of the picture, and self-reported camouflaging is associated with worse mental health outcomes across much of the literature (Cook et al., 2021). Those consequences are often invisible in an appointment that ends before they start.

Ask what the conclusion rested on. If a result doesn’t fit, the specific question is more useful than the general objection: which sources of information were used, whether developmental history was gathered, and whether an informant or documentary evidence was part of it. A second opinion is a legitimate next step, and clinicians vary in how familiar they are with adult presentation.

If you want the fuller picture of what masking is and what it costs day to day, our guide to autistic masking goes deeper than this page does. To get your own history and examples in order before an appointment, the NeuroDiversion AuDHD self-reflection is built for that step. It isn’t a diagnosis, and nothing here replaces a clinician.