One hundred and ten consecutive patients admitted to an acute psychogeriatric unit were assessed with the Geriatric Mental State Schedule, assigned a DSM-111-R diagnosis and their length of stay in hospital and discharge destination noted. Forty-two per cent of patients had organic disorders (mainly dementia), 32% had affective disorders (mainly depression) and 23% had schizophrenia or a related disorder. The median length of stay was 29 days and bore little relation to diagnosis, age, sex or cognitive function. Patients with an organic mental disorder were more likely to enter long-term institutional care, but 78% of patients were able to return to the same level of accommodation (own home, hostel or nursing home) that they had occupied prior to admission. On a four-point non-blind outcome scale, 86% of patients were rated as improved at discharge. KEY wow-psychiatric disorder; prevalence; elderly; psychogeriatric unit In recent years specialist psychiatric services for the elderly have proliferated in several developed countries, but randomized clinical trials to determine their effectiveness have not been performed (Cole, 1988). Epidemiological data and population projections alone will not convince health planners of the need for maintenance o r expansion of services unless formal evaluations demonstrate their worth (Draper, 1990). Yet even straightforward published descriptions of the work of specialist acute psychogeriatric units are few in number and flawed in design. We have traced six such studies published in the last 16 years. Two studies from Australia (Gil-