A survey of people with severe intellectual disability and the most severe challenging behaviour in Wales identified five adults living in family homes, 17 in new specialist community housing and 19 in traditional services. With the omission of two people from the latter group and with a restricted collection of data for people living in the family home, the present study explored service input, outcome and costs across the three setting types. Process and outcome indicators for the family home group, who received little service input, were better than those for the traditional service group, although less good than those for the community house group. The specialist community home model produced significant gains over the traditional services in virtually all areas. Across the residential data set as a whole, there was no association between staff:resident ratios and severity of disability or between costs and severity of disability. This was largely true of the service types separately. There was a relationship between costs and service quality. However, this association was underpinned by gross differences between community houses and traditional settings. Costs, processes and outcomes ceased to be related when the two residential types were considered separately. Although higher costs of new community services compared to traditional services may be set against improved outcomes, high costs within the former could not be related to benefit. Outcome indicators were generally related to each other, suggesting that high quality in one sense was matched by high quality in other senses. Outcome was significantly associated with the ability of residents. Outcome indicators also tended to be related to observed staff performance, which was independent of resident ability. Therefore, outcome may be considered as dually determined by differences in resident ability and in what staff did.