This paper presents data from a longitudinal collaborative study of The Care in the Community Sample (Cambrdige, Hayes, Knapp, Gould, & Fenyo, 1994; Cambridge et al., 2001; Knapp et al., 1992). The aim of the study was to investigate how social impairment changes are related to the move from institutional to community care using some preliminary analysis of the above data. A measure of social impairment using the Skills and Behavior Interview from the most recent follow-up of this cohort was found to be consistent with Wing's definition of social impairment, when applied to the cohort 12 years after deinsitutionalization. This measure was then used to retrospectively identify social impairment in the same sample (of approximately 250 people) at baseline (in the institution), at 1 year and at 5 years in the community. Prevalence data pointed to high levels of social impairment in the sample (50.1%) in institutions but the decrease to 39.8% after 1 year in the community was not significant (although conversation and social mixing and initiation of conversation and social interaction did improve over time). Once in the community, social impairment in general did not change over time, although there was a significant decrease in conversation and social mixing, non-verbal communication and initiation of conversation and social interaction. These results are compared to other research findings and the implications and limitations of the study discussed.