Severity of psychiatric illness was assessed using standardised clinical and social measures in 69 in-patients and 41 day patients admitted consecutively from the community. Day and in-patients differed little in terms of psychiatric symptoms and social disability, especially if compulsory admissions were excluded. Protection of self or others was a common reason for in-patient admission given by clinicians, who were otherwise prepared to treat seriously ill patients in the day hospital. Very few of the day patients had to be transferred to the in-patient facility, and at three months and one year the two groups showed similar improvements. It is concluded that day treatment is feasible for some seriously ill psychiatric patients, but a random-allocation study is required to assess more completely the efficacy of day treatment, and define the characteristics of those who require in-patient admission.