Information was collected on the social networks of  adults with mental retardation receiving different types of residential support. The results indicated that: () the reported median size of the participants’ social networks (excluding staff) was two people; () % of participants were reported to have a member of staff, % a member of their family, % another person with mental retardation and % a person who did not fit into any of these categories in their social network; and () variation in the size and composition of participants’ social networks was associated with a range of variables including the personal characteristics of residents (i.e. age, autism, ability and challenging behaviour), the type of previous and current accommodation, staffing ratios, institutional climate, and the implementation of ‘active support’.