The AQ-10 is a ten-item self-report screen for autistic traits in adults with average or above-average intellectual ability. Each item is a statement about everyday social communication, attention switching, attention to detail and imagination, answered on a four-point agree–disagree format that is collapsed to a binary score: one point for each autism-consistent response, giving a total from 0 to 10.
It is a screening instrument, not a diagnostic one. A score at or above threshold indicates that a specialist autism assessment is worth considering; it says nothing definitive about whether the person is autistic. Nor is it a trait severity measure - recent psychometric work cautions against using the total score as a continuous measure of autistic traits in the general population.
02 - Origin & purpose
Where it comes from.
The AQ-10 was derived by Allison, Auyeung and Baron-Cohen at the Autism Research Centre, University of Cambridge, and published in 2012 in the Journal of the American Academy of Child and Adolescent Psychiatry. Working from the full 50-item Autism Spectrum Quotient (AQ-50), the authors used discrimination-index analysis in around 1,000 diagnosed cases and 3,000 controls to select the ten items that best separated the groups, producing adult, adolescent and child versions.
The explicit purpose was a "red flag" instrument brief enough for routine use in primary care and general clinics - somewhere a GP or psychiatrist could quickly decide whether a specialist referral is warranted. In England, NICE guideline CG142 recommends considering the AQ-10 in adults with possible autism who do not have moderate or severe learning disability when deciding on referral for assessment.
03 - Scoring & cutoffs
How scoring works.
Each item scores 1 when the response is autism-consistent (for some items "agree" responses, for others "disagree" - the calculator above handles the keying). Total 0–10. The published referral threshold is 6 or more: at this cut-point the adult version showed sensitivity 0.88 and specificity 0.91 in the derivation sample. Note that the original paper's abstract phrases the recommendation as "more than 6 out of 10", while the widely used ARC scoring sheet and NICE apply ≥6 - state the threshold you use in your documentation. Scores just below threshold in a person with a convincing developmental history should not, on their own, close the door on referral.
Score
Severity
Interpretation
0–5
Below threshold
Score below the referral threshold for autism assessment.
6–10
At or above threshold
A score of 6 or more suggests considering referral for a specialist autism assessment.
04 - Validation evidence
How well it performs.
In the 2012 derivation study (case-control, ~1,000 adults with autism diagnoses vs ~3,000 controls), the adult AQ-10 at cut-point 6 showed sensitivity 0.88, specificity 0.91 and positive predictive value 0.85. Booth and colleagues (2013) independently evaluated the AQ-10 in 149 adults with ASD and 134 controls and found screening performance essentially equivalent to the full AQ-50, supporting its use as a brief substitute.
The main caveat comes from general-population work: Taylor, Livingston and colleagues (2020) found internal consistency below conventional standards (Cronbach's alpha and McDonald's omega < .70, average inter-item correlation < .20), meaning the AQ-10 behaves acceptably as a clinical referral screen but poorly as a trait measure in non-clinical samples. Remember that case-control designs typically inflate accuracy relative to real-world consecutive referrals.
0.88
SENSITIVITY (≥6, ADULT DERIVATION)
0.91
SPECIFICITY (≥6, ADULT DERIVATION)
0.85
PPV (≥6, ADULT DERIVATION)
α < .70
INTERNAL CONSISTENCY, GENERAL POPULATION (TAYLOR 2020)
05 - How it compares
How it compares to the alternatives.
Instrument
Items
Time
When to reach for it
AQ-10
10
~2 min
The brief referral screen itself: ten items, threshold ≥6, intended for adults without learning disability in primary care and general psychiatry.
AQ-50
50
~10 min
The parent instrument; richer trait coverage when you want subscale detail. Free from the Autism Research Centre for non-profit use.
RAADS-R
80
~30 min
Deeper adult measure including developmental history and camouflaged presentations; heavier burden.
Parent/caregiver-report about developmental history, mainly for children; licensed (WPS).
ADOS-2
-
40–60 min
Observation-based, semi-structured diagnostic-assessment component, clinician-administered and licensed - not a screen, but what a positive AQ-10 leads towards.
06 - When to use it
Right tool, wrong tool.
The AQ-10 is intended as a referral screen for adults with possible autism who do not have moderate or severe learning disability. It suits primary care and general psychiatry, where the question is whether a specialist autism assessment is warranted, not what the diagnosis is.
A positive screen should lead to specialist assessment including developmental history; a negative screen does not close the question when the history is convincing.
Reach for it when
-Deciding whether to refer an adult without learning disability for specialist autism assessment.
-A quick structured prompt in primary care or general psychiatry.
-Adding structure to a conversation the patient has already opened about possible autism.
Reach for something else when
-Diagnosing autism - no self-report scale can.
-Measuring autistic-trait severity or change over time.
-Screening adults with moderate-to-severe learning disability.
-Children and adolescents - use the age-appropriate AQ-10 versions or the SCQ.
-Ruling out autism when the developmental history is convincing; sensitivity 0.88 still misses roughly one in eight.
07 - Confidence & precision
Reading the score with care.
No standard error of measurement or meaningful-change threshold is established for the AQ-10, and none should be expected: it is a binary-keyed referral screen, not a measurement instrument. The low internal consistency reported in general-population samples means single-point differences are noise; treat the result as above/below threshold rather than a quantity. If you need to track traits or symptoms over time, use a fuller instrument.
08 - Limitations
What it cannot tell you.
Derivation and most validation are case-control, which inflates accuracy versus consecutive clinical samples. Self-report requires insight into one's own social behaviour, which can be reduced in autism and can also be shaped by camouflaging, especially in women.
Internal consistency is poor when the scale is used as a trait measure in the general population (Taylor et al. 2020). It is validated in adults with IQ in the average range and above, not in learning disability. Item keying is mixed, so hand-scoring errors are common without the scoring sheet. Elevated scores also occur in social anxiety, ADHD and psychosis, so the threshold indicates assessment, not diagnosis.
See how Aisel removes friction where it costs most. A 20-minute walkthrough tailored to your clinic.
We value your privacy
We use cookies to analyse site usage and improve your experience. Analytics and embedded media (e.g. YouTube) only load if you accept. Read our cookie policy.