Ambulatory care for subjects with severe mental problems has been clearly shown to be a valid alternative to hospitalisation. However, very few studies have considered the fate of patients over several years. Ambulatory care services are often experimental set-ups, for small groups, and their impact on subsequent treatment has only been assessed over the first few months of treatment. The value of developing this practice therefore remains unclear. We investigated the possible consequences of generalising ambulatory care services by a mobile crisis intervention team (ERIC) to all requests for the first hospitalisation in a psychiatric department. The principal aim was to determine whether systemic intervention by the crisis intervention team could provide a true alternative to hospitalisation. We also investigated whether problem-resolving approaches and ambulatory care led, in the long term, to fewer prolonged or repeated periods of hospitalisation than practices in which hospitalisation was considered as an ordinary solution. We carried out a prospective, comparative, cohort study over a 5-year period beginning with the creation of ERIC by one of the hospital departments. All patients arriving at this department for the first time were offered immediate ambulatory care by this team for 1 month. Their hospitalisation record (duration of hospital stay, number of days in hospital) was compared with that of subjects hospitalised in the same conditions but in other departments of the hospital. This study included most of the subjects referred for the first time to the psychiatric hospital, in our department. Regardless of their diagnosis, intensive follow-up at home, based on systemic crisis intervention work, was found to be an effective and well-accepted alternative to hospitalisation. Indeed, a highly significant immediate decrease in both the number of admissions and the duration of hospital stay was observed for the experimental group, with no subsequent increase in the number of days of hospitalisation. From the second year onwards, the use of hospitalisation did not seem to be influenced by the type of care initially given to the patient. Rehospitalisation was rare in both groups. One third of the patients in the experimental group benefited from another intervention of the ambulatory emergency team from the second year onwards, highlighting the value placed on this type of care by the patients and their families. Our results support the development of ambulatory crisis intervention services, including those from psychiatric hospitals. Clinical studies following the treatment paths of patients in a more exhaustive manner would almost certainly distinguish more precisely between the "natural" course of the disease and the impact of the care provided. In any case, the prevention of hospitalisation must be based as much on a possible alternative at the time of the crisis as on subsequent access to ambulatory care.