This report describes a prospective study of a geriatric day hospital. All 334 patients who attended during a six-month period were included with follow-up for a minimum of six months for each patient. One third were patients who had been discharged from the acute geriatric ward and most of the remainder were direct referrals from general practitioners. Transport was provided in 92% of cases by the Ambulance Service using a single-person crew in two-thirds and a driver plus attendant in the remainder. Only 8% came by private transport (family car). Patients were classified as ‘once-only’ attenders (22%), other new patients (37%), re-referrals (11%), current attenders (19%) and 11.1% were patients for whom a positive decision had been made to continue their day hospital attendance until death or admission to an institution (so-called ‘chronic attenders’). This day hospital is an integral part of a comprehensive geriatric service which relies upon early referral and immediate response to avoid waiting lists both for in-patient and day hospital care. The services used by day patients are given with the proportion of patients receiving each service. The ecology of the day hospital was demonstrated to be more or less in balance. It is especially notable that the largest proportion of patients were discharged (from 41% to 63% of different patient categories) and there was no accumulation of chronic attenders. Of the 260 patients (excluding only the 74 once-only patients), only 25 were in geriatric accommodation and 12 in residential homes after a minimum of six months' follow-up. Forty-four patients had died. It is postulated that, in order to make an optimum contribution to an active geriatric service, the day hospital should be sited alongside the acute admission unit and it must be visited by senior doctors from the service at least as often as they visit the acute wards. Adequate nursing, rehabilitation and other paramedical staff must be provided as well as suitable transport. Decisions on patient management and future care should be based upon regular case conferences. Each day hospital ought to state its general policy and there should be declared goals for each patient. Once these goals have been achieved, the patient should move on to some other form of care (apart from designated chronic attenders who probably should not constitute more than 10–15% of day patients).