This thesis is based on an ethnographic study of the way in which care in the community was delivered in two privately run specialist hostels for homeless people. One was for people recovering from severe mental illness, the other for people with addiction to alcohol. Medication was a significant part of the intended therapy and its use and management was rooted in the wider culture of the two hostels. The study aimed to understand what care in the community meant for the lives of the people who lived in these hostels. I set out to discover what kind of care such hostels were providing and to evaluate the provision against the aspirations expressed in the community care legislation and guidelines. The main aims of care in the community included: a reduction in the numbers in hospitals, a reduction in the negative effects of institutionalisation, improving standards of accommodation, humanising the care, developing more psycho-therapeutic approaches, recognising the civil rights of vulnerable groups, directing the right services to the right people who needed them, develop systems that ensured the CO-ordination and continuity of care and involving service users in formulating, implementing and reviewing care. The model of care espoused in both hostels was skills-based rehabilitation towards independent living but analYSis of the data evoked enduring aspects of the culture of care that echoed life in Goffman's (1961) "Asylums". While much had changed in the physical environment and in the expressed objectives by staff data collected from participant observation and interviewing fell broadly into two categories: staff practices that worked in harmony with residents' patterns of needs and those that repeatedly resulted in control and conflict. Medication use had to be seen in this context. Analogies could be drawn with what Goffman (1961) conceptualised as "privileges" and "mortifications". Further similarities with Goffman's study were apparent in the way hostel staff rationalised their actions. Goffman's approach to "career" and "stigma" were also used in an 11 analysis of discourses in the social life of the two hostels. I drew on Foucault's (1967, 1982) work on discourse to make links with power and knowledge structures in relation to mental illness. Two discourses were evident; one based on risk-as-danger and stigma and the other on liberal independence and integration. Risk-as-danger discourses were dominant. This discursive framework also links to wider policy and practice, including legislation and guidelines. The notion of surveillance (Foucault 1977) adds further insight into the issues of danger, control and independence. I argue that these continuities reflect a culture of control that has transcended the closure of the total institution. I locate the causes in social stigma, the unaccountable autonomy of hostel managers, the lack of staff training and a model of rehabilitation that focuses on basic domestic tasks, rather than specific therapeutic interventions appropriate for individual recovery. The conflict around medication, the mistrust of care givers and the breaks in the continuity of care when residents moved to independent living tended to lead to relapse and readmission to hospital. Although residents benefited greatly from the social relations they developed with one another.