Autistic people are dramatically overrepresented in homeless populations by at least 6 times the general population rate, and by some estimates considerably more. Research screening homeless outreach caseloads found 12.3% met DSM-5 autism criteria, against a general population prevalence of roughly 1-2%. A 2022 systematic review spanning 17 studies found prevalence estimates of autistic traits in homeless communities ranging as high as 50%. The most defensible midpoint sits around 20-25%.
Applied to HUD’s 2024 point-in-time count of 771,480 homeless people, that means somewhere between 95,000 and 193,000 people experiencing homelessness are likely autistic and at the upper-bound estimate, as many as 385,000. At the undiagnosis rates we know apply to this age group, 75% for adults 20-49, over 90% for those 50 and older, the overwhelming majority of those people have never received a diagnosis. Only a minority of autistic homeless individuals had a clinical autism diagnosis at the point they first became homeless. The system that was supposed to catch them failed before they ever reached the street.
This is not a peripheral finding. It means homelessness services are operating in near-total diagnostic blindness toward a neurological profile that is 6 to 25 times more common in their population than in the general public. Standard intake processes, shelter environments, and case management protocols are built around social and communicative norms that autistic people particularly undiagnosed autistic people, who have no framework for understanding their own responses are structurally unable to meet.
Sensory overwhelm in shelter settings, difficulty with unstructured social demands, alexithymia that reads as non-compliance or disengagement, rigid routines disrupted by shelter life: these are not behavioral failures. They are the predictable consequences of placing an unrecognized neurological minority inside systems designed for a different nervous system, with no accommodation and no identification.
The result is that autistic people are not just more likely to become homeless, they are more likely to stay homeless longer, to be excluded from or exit services prematurely, and to cycle through crisis systems repeatedly without any of those systems naming what is actually happening. The harm is not incidental. It is structural, it is ongoing, and it is operating at a scale of tens to hundreds of thousands of people.
The claim is not that autism explains all homelessness. The claim is that homelessness services are systematically blind to a neurological profile massively overrepresented in their population, and that blindness is causing compounding, measurable harm at scale; harm that better identification, neurologically-informed housing design, and community-integrated support could meaningfully interrupt.
NeuroHomes 