Trained clinicians interviewed 346 people who were representative of those aged 15 to 60 years staying in crisis accommodation centres for the homeless and cheap single-room accommodations in inner Melbourne. The interviewers used a standardised diagnostic instrument, the structured clinical interview for DSM III-R. (Diagnostic and statistical manual of mental disorders - revised), to diagnose a range of severe mental disorders, including psychotic, affective, and substance-related disorders. Almost half the people interviewed received diagnoses of current disorders and over 70% received diagnoses of lifetime disorders. The prevalences of lifetime and current disorders in all categories were as high in young as in older men. Only small numbers of women were seen. The relatively high prevalence of disorder in younger men may be related to selective factors in the survey, to a cohort effect, or to recovery or death of older men with a history of mental disorder. From a practical point of view the important issue is the effect of varying systems of mental health care, and of welfare and housing policies, on the course and outcome of the various disorders, and on the likelihood of individuals living impoverished and disaffiliated lives.