Police, courts and prison

On the best comparison evidence, autistic people are not over-represented among those charged with or convicted of crimes. In a New Zealand birth cohort that followed 1,197 autistic and 147,879 non-autistic young people to age 25, autistic young people had lower rates of being proceeded against by police (37.6% lower), charged (39.2% lower) and convicted (43.1% lower), and the overall risk of imprisonment did not differ between the two groups.¹ Among those who were charged, autistic young people were more likely to face a serious or violent charge, and more likely to be imprisoned once charged (13%, against 8%).¹ A 2014 systematic review reached the same broad conclusion, and a 2023 update found that the studies published since still carry too many methodological limits for direct comparison.^2,3

The UK Millennium Cohort Study, which followed 11,859 young people, gives a more detailed picture. By age 14, 16.5% of the whole sample had been stopped by police, 9.3% had been given a warning and 1.3% had been arrested. Autistic adolescents were about twice as likely to report a warning before other factors were taken into account, but the difference was no longer significant once delinquency, poverty and other factors were adjusted for, and in that adjusted comparison they were less likely to have been stopped. There was no difference in arrests.⁴

A specialist adult autism service in London found that 23% of the adults it diagnosed had earlier contact with the justice system as a potential suspect, against 32% of the adults it assessed but did not diagnose, and no offense category where the autistic group was higher. Being male, and having ADHD or psychosis, were associated with higher odds of that contact.⁵ The comparison group was other people referred to the same service.

The picture varies by subgroup and offense type. In a Danish register study of 313 former child psychiatric in-patients with a pervasive developmental disorder, conviction rates ranged from 0.9% in childhood autism to 18.4% in Asperger's syndrome, against about 15% to 20% in matched controls.⁶ In a South Carolina records study, young people with autism were charged with more crimes against people and fewer against property, and were more likely to be diverted and less likely to be prosecuted.⁷ A 2024 review of the wider literature found the studies too varied and weak to support a definitive link between autism and criminal conduct.⁸ Where autistic people do offend, one study found that conduct problems separated autistic offenders from autistic non-offenders, while mental health difficulties separated them only from the non-autistic comparison groups.⁹

Most autistic people are never arrested, but those who are report being treated less well than non-autistic people. In an online survey of 145 autistic and 116 non-autistic adults who had been arrested, the autistic participants were less satisfied than the others with how police treated them at arrest, in custody and during questioning.¹⁰ They were also far more likely to say they had not been given an appropriate adult they believed they needed: 42%, against 13% of the others.¹⁰ Among the diagnosed autistic participants, 82% described meltdowns and 62% described shutdowns that they linked to their involvement with the justice system.¹⁰

In a survey of lawyers about 93 cases, 75% of the autistic clients had not been given reasonable adjustments and only 43% had been offered an appropriate adult, even though they already had a diagnosis.¹¹ Screening at six London police custody centers flagged 5% of arrestees who had no existing diagnosis as warranting further assessment for autism.¹² In interviews with autistic adults and their parents, autistic characteristics were described as shaping 92.3% of police interactions, most often by affecting how the interaction went.¹³

As of October 2026, police in England and Wales are public authorities under the Equality Act 2010 and must make reasonable adjustments for disabled people, autism included, and PACE Code C requires an appropriate adult for a vulnerable person in custody.^14,15

How an autistic person comes across in an interview can shape how they are judged. In two experiments, observers watched recorded interviews with autistic and non-autistic suspects and rated the autistic male suspects, and autistic suspects who hesitated more in speech, more harshly. The observers' own diagnostic status made no difference, and the pattern followed expectancy violations: ratings fell when a suspect behaved in a way the observer did not expect.¹⁶

A separate study had autistic and non-autistic adults take part in a simulated online chatroom about illegal hacking, then interviewed them by video. A diagnosis of autism had a strong effect on a person's overall behavior, but only a few individual behaviors differed between the two groups, and no behavior was shown consistently by everyone. The differences that did show up, difficulty with figurative language, longer speech hesitations and more variable intonation, are the kinds that can count against a person in a high-stakes interview. The authors note that the setting was not face to face, and that the behavioral differences were more noticeable in the men than the women.¹⁷

Legal professionals in England and Wales, asked about their work with autistic people in criminal courts, identified training, understanding and adjustments as the main problems. The autistic adults and parents surveyed alongside them were largely dissatisfied with their experience.¹⁸

Giving a jury information about autism may change how it reads a defendant, at least in the setting studied. In one experiment, 160 jury-eligible participants read a single case about a man brought to police attention for suspicious and aggressive behavior; half were told that he was autistic and given background information. Those participants rated him more honest and likeable, blamed him less and returned fewer guilty verdicts than the participants who were given no label. The effect may depend on how the defendant behaved before the label is given.¹⁹

Courts can also be adapted. In the first case of its kind in New South Wales, a judge appointed a communication intermediary for an autistic adult defendant on a sexual assault charge. Lawyers followed the intermediary's recommendations and asked mostly simple questions, and the defendant gave his evidence with confidence, using visual aids and asking for clarification when he needed it.²⁰

Sentences may fall unevenly. In Australian data, autistic offenders received longer sentences than the national averages for similar offenses, with a significant difference for sexual assault and none for murder, manslaughter or assault. That study compared autistic offenders with national sentencing figures, not with matched non-autistic defendants.²¹ The mock-juror experiment above found the opposite direction, with more lenient sentencing when a label was given.¹⁹

Before a trial, the law asks whether a defendant can take part. As of October 2026, fitness to plead in England and Wales still rests on criteria set out in 1836 in R v Pritchard, which the Law Commission recommended replacing in January 2016 with a test based on a defendant's decision-making capacity and ability to take part effectively in their trial; that reform has not been enacted.²² In the United States, competence to stand trial follows the 1960 Supreme Court decision in Dusky v. United States: whether the defendant has enough present ability to consult with a lawyer with a reasonable degree of rational understanding, and a rational and factual understanding of the proceedings.²³

Researchers have built a standardized instrument to assess fitness to plead, after finding that the current process is subjective and that findings of unfitness are rare enough to cause professional unease.²⁴ How autism itself bears on responsibility is unsettled. In a Swedish study of 20 forensic evaluations of violent offenders with autism from 2016 to 2018, evaluators gave more weight toward a finding of severe mental disorder to very low psychosocial functioning, a fixation on violence and rigid coping strategies.²⁵ Insanity-defense criteria differ between jurisdictions, and people with a neurodevelopmental disorder may be competent to stand trial yet still meet the criteria for an insanity defense, according to one review.²⁶ Clinician-researchers writing about New South Wales, Norway and England and Wales describe uneven screening and wide variation in what happens to defendants with intellectual disability or autism, and say all three places had some process for supporting vulnerable defendants.²⁷

Autism is recorded more often in prisons than in the general population, though the estimates vary. Across the 10 English women's prisons that released exact counts, 141 of 2,949 women were recorded as autistic, a rate of 4.78%, about 13.7 times the general-population rate for autistic women.²⁸ A systematic review found estimates of autism in the justice system ranging from about 1% to 60%, depending on the setting and how autism was identified.²⁹ A 2016 article in The BMJ argued that autism is underdiagnosed in prisoners.³⁰

The higher recorded rates sit alongside the lower rates of charging. Researchers suggest autistic people may be more likely to be caught, or that a lack of autism-sensitive programs and assessments may lead to longer stays once inside.²⁸

Prison is hard to adapt. An interactive autism-awareness training for prison staff raised their knowledge of autism but did not change their stigma toward autistic people, compared with staff who had not taken it.³¹ A systematic review of interventions for offending behavior in autistic adults without intellectual disability found only seven studies, with ten participants between them, all male, and judged the interventions largely inadequate.³²

Officers who take autism training know more about autism afterward and feel more confident. In a study of 259 law enforcement officers, scores on both rose after an autism presentation, whether or not it was combined with virtual-reality or in-person simulation; adding the simulation gave no extra gain.³³ A study of first responders found the same broad pattern across roles, though officers began with higher self-efficacy and improved less.³⁴ Both studies measured change in the responders themselves.

No study found as of October 2026 has measured whether police autism training changes what happens to an autistic person in a real encounter.

The trials that exist put the work on the autistic person. In a randomized trial of 47 autistic participants aged 12 to 60, practice in a virtual-reality police-safety module reduced fidgeting during live practice interactions with officers significantly more than a video-modeling program did, and behavior within the virtual-reality group improved in follow-up tests.³⁵

Autistic writers describe the police as a place where being autistic, and being Black, can be dangerous. Jules Edwards tells the story of Kobe Heisler, a biracial autistic young man killed by police after his grandparents called 911 during a meltdown, and writes that law enforcement is repeatedly called on dysregulated autistic people. She argues that "the first goal of law enforcement is to gain control", and that common autistic traits, a slower response, echolalia, motor differences, going quiet, are read by officers as "pre-attack indicators". Edwards also argues against schemes that issue identification tags or cards, on the grounds that they mark disabled people as targets.^c1

Melanie Yergeau names the targeting of autistic people of color by law enforcement "being autistic while black", and recounts the case of Reginald "Neli" Latson, arrested in 2010 and later conditionally pardoned. Eric Garcia writes about the police killings of the autistic Black young men Stephon Watts and Ricardo Hayes, and about racialized autism, the reading of autistic behavior through race.^c2,c3

Devon Price places policing inside a wider argument for abolition, writing that disabled people are at high risk of being shot by police and that Black and brown autistic people face elevated risk of police violence and incarceration.^c4 Luke Beardon writes that autistic people may take what they are told as truth, obey instructions and see no alternative, which can lead them into criminal behavior, and that the justice system mistranslates autistic behavior such as avoiding eye contact, curtness or a flat tone as guilt or a lack of remorse. Price gives the case of Paul Modrowski, sentenced to life by a judge who said he "looked like a murderer" because he avoided eye contact and sat still in court.^c5,c6