Feasibility of day-treatment for patients accepted for admission to a psychiatric hospital was tested by means of a randomised experiment. In search of patient characteristics indicating suitability for day treatment, specific and general social role dysfunction were measured at the time of admission, covering eight domains of social role performance. Analysis of variance shows that experimental and control patients do not differ in social role performance before admission. Dysfunction in general and in the specific domains is severe, particularly in the occupational, social, partner and household roles. Day treatment appeared feasible for 36% of the experimental patients, in spite of the fact that these patients' social dysfunction is no less extended or severe, save in a restricted number of domains, namely the management of financial affairs, the rearing of children and the relationship with the partner.