We have developed a protocol to identify unnecessary days of hospitalisation in the Department of Neurology of the Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland. Seventy-four parameters (medical, social, type of investigation and treatment, degree of disability and of dependence) potentially associated with the length of stay were studied prospectively in 511 nonselected patients consecutively admitted to the Department over a period of 5 months. Each day spent on the wards was analyzed on a day-to-day basis and was classified into one of two groups: those due to a medical reason (4,700 hospital days), and those due to a nonmedical reason (1,184 days). These delays resulted chiefly from difficulties in obtaining laboratory investigations, especially in patients who were not disabled and who had been admitted for investigations (3.8% of hospital days, compared to 1.5% for patients with severe or total dependence) or from awaiting transfer to either another department or a nursing home. This second cause of delay markedly increased the length of stay of patients with severe or total dependence and with limited mobility on the first day (26.0 days, compared with 7.4 days for nondisabled patients) and, above all, on the last day spent in our Department (27.0 days, compared with 8.0 days). The ongoing analysis of these data may provide information as to which parameters could be influenced by neurologists in order to reduce the length of stay in hospital.