Objectives – The evaluation matrix recently proposed by Tansella and Thornicroft suggests that the field of social and epidemiological psychiatry has focussed more on the individual/patient level of mental health care services than the system level. Moreover, phenomena such as deinstitutionalisation have been examined more as clinical events than as social ones. The aims here are to deepen our understanding of deinstitutionalisation, particularly as regards the downsizing/closure and role of psychiatric hospitals. Methods – I begin by reviewing the manifest and latent functions of psychiatric hospitals. This is followed by a discussion of how these functions must be met by any comprehensive community-oriented system of mental health care for severely mentally ill patients. Also, in order to reframe the downsizing/closure of psychiatric hospitals as a social event for the field of social psychiatry and psychiatric epidemiology, I posit that the process of deinstitutionalisation is driven today by the same forces that were present at the outset of the movement. Results – I review four recent series of studies addressing primarily the outcomes, but also other aspects, of the downsizing/closure of psychiatric hospitals, with a view to illustrating the methods used, the results obtained and the blind angles missed in this research. Conclusions – Lessons are drawn on how to fill certain vacant cells of the matrix.