Background: There is little evidence to guide the commissioning of residential provision for adults with autistic spectrum disorder (ASD) in the UK. We aim to explore the degree and impact of diagnostic congregation among adults with intellectual disabilities (ID) and ASD living in staffed housing. Methods: One hundred and fifty-seven adults with intellectual disabilities from a sample of 424 in staffed housing were assessed as having the triad of impairments characteristic of ASD. They lived in 88 houses: 26 were non-congregate (40% or fewer residents had the triad) and 50 congregate (60% or more had the triad); 12 with intermediate groupings were eliminated. Non-congregate and congregate groups were compared on age, gender, adaptive and challenging behaviour, house size, staff per resident and various measures of quality of care and quality of outcome. Comparisons were repeated for Adaptive Behavior Scale (ABS)-matched, congregate and non-congregate subsamples. Results: Non-congregate settings were larger, had lower staff per resident and more individualised social milieus. Groups were similar in age and gender but the non-congregate group had non-significantly higher ABS scores. The non-congregate group did more social, community and household activities. After matching for ABS, these outcome differences ceased to be significant. Non-congregate settings were significantly larger and had significantly more organised working methods. Conclusions: The findings are consistent with other research that finds few advantages to diagnostic grouping. (Contains 2 tables.)