Autistic women and girls are identified later and less often than autistic boys, and the gap is not fully explained by there being fewer of them. Loomes and colleagues pooled the prevalence studies and found the ratio narrows substantially when researchers screen a population directly rather than counting who got referred, which means a real share of autistic girls are never sent for assessment at all. [loomes-2017-ratio] This article covers what presents differently, why the diagnostic picture missed it, and what changes with a late diagnosis.
Why the picture was drawn wrong
The behaviours clinicians are trained to recognise came from the samples the criteria were developed on, and those samples were mostly boys. That is a fact about the history of the research rather than an accusation, and its consequence is mechanical: a presentation that differs from the reference picture reads as evidence against autism instead of as autism presenting differently.
Lai and colleagues set out the problem directly, arguing that sex and gender differences in how autism appears have been under-studied relative to how much they affect who gets identified. [lai-2015-sexgender]
Referral compounds it. Assessment starts when an adult raises a concern, adults raise concerns about children who are disruptive, and the modal missed autistic girl is compliant, quiet, anxious and working extremely hard to keep up. Our guide to autism covers what the condition involves in general, and our guide to child mental health covers that same under-referral of quiet children across conditions.
What actually differs
| The reference picture | How it often presents in girls and women | |
|---|---|---|
| Friendships | Few or no friendships sought | Intense friendships, often one at a time, hard to sustain |
| Special interests | Content marks it out (timetables, systems) | Same intensity, socially ordinary content (a band, animals, an author) |
| Stimming | Visible, large movements | Small and concealed: hair twirling, skin picking, movement inside a pocket |
| Social difficulty | Apparent immediately | Hidden by rehearsed scripts and copied behaviour |
| Sensory sensitivity | Reported as sensory | Reported as anxiety, or not reported at all |
| Where it shows | Consistently, across settings | Held together in public, collapse at home |
The last row is the one families find hardest to believe, because the school reports nothing wrong. A child who is holding themselves together all day is not fine; they are spending everything they have on looking fine, and the bill arrives at four in the afternoon.
Camouflaging, and what it costs
Camouflaging is the effortful work of appearing non-autistic, and Hull and colleagues documented what autistic adults describe doing: rehearsing conversations in advance, copying other people’s expressions and phrasing, forcing eye contact at intervals, suppressing stimming, and monitoring themselves continuously. [hull-2017-camouflaging]
It succeeds, which is precisely the trouble. It removes the evidence a clinician would use, so it delays diagnosis by years. And it is sustained effortful performance during exactly the situations other people find restful, so it produces exhaustion that looks unexplained from outside.
- Masking discomfort and stimming 35% of reported strategies
- Copying others deliberately 30% of reported strategies
- Rehearsing scripts in advance 25% of reported strategies
- Other strategies 10% of reported strategies
The strategy categories described by Hull and colleagues (2017). Values illustrate the relative prominence of each rather than reporting measured proportions.
Stopping is harder than it sounds. The mask is usually built in childhood without a decision being made, so unmasking is less like dropping a habit and more like learning which parts were ever chosen.
What gets diagnosed instead
Anxiety and depression are the usual first labels, and they are usually genuinely present. The difficulty is that treating them without knowing what generates them tends to produce partial results, and the person concludes that treatment does not work on them.
Eating disorders, ADHD and borderline personality disorder are the other common first diagnoses. That last one is worth naming: emotional intensity plus unstable friendships plus a history of not fitting can look like a personality disorder if nobody asks whether the pattern was there at six. Our guides to ADHD in women and BPD cover the two most often confused with it.
The distinguishing question is developmental. A lifelong, constant difference is a different thing from an episodic condition, whatever the surface resemblance.
Does this pattern sound familiar?
For yourself or for a daughter. This is not a test, it cannot diagnose anything, and no screener on this site assesses autism. It is a prompt for what to describe if you seek an assessment.
0 of 7 ticked
This is the shape that gets missed. Bring specific childhood examples and specific present-day costs, because a short appointment with somebody who camouflages well is exactly the situation in which this is overlooked.
Enough of the pattern to be worth reading more about, and worth raising if you already see someone. Write down childhood examples, since assessment rests heavily on developmental history.
Little of this is present. If something is still difficult, it is worth pursuing on its own terms rather than through this lens.
No autism screener is published on this site. The anxiety screener is here because anxiety is the most common thing that accumulates on top, and it is treatable now regardless of what else is going on.
Late diagnosis
Most women identified in adulthood describe the explanation as the valuable part, ahead of anything it unlocked administratively. It reframes a large amount of accumulated evidence at once: the exhaustion was real, the sensory intolerance was not fussiness, and the effort other people did not appear to be making genuinely was not being made.
It also changes what help makes sense. Pacing, sensory accommodations and permission to socialise less are interventions that anxiety treatment on its own does not supply.
One thing worth stating plainly rather than leaving in the background: Cassidy and colleagues found markers of suicidality elevated in autistic adults, with camouflaging and unmet support needs among the factors associated with it. [cassidy-2018-risk] That is an argument for taking distress in this group seriously and early, not for treating autism as a tragedy.
When to seek help
Speak to a doctor if the pattern above fits and is costing you work, study, relationships or health, and ask specifically about an autism assessment rather than describing only the anxiety on top, since the anxiety is what usually gets treated in isolation. Waiting lists are long in most places, so it is worth asking what the route is early. Seek help sooner if low mood has been present for weeks, if exhaustion has stopped you functioning, or if you are having thoughts of harming yourself.
If you are in crisis, contact your local emergency services or a crisis helpline.