Various studies have shown that adverse drug events (ADE) are a substantial cause of hospital admissions. However, little is known about the incidence and severity of ADE resulting in hospital visits. To address this issue, we conducted a prospective survey in 10 primary care and emergency departments of French public hospitals. This study was performed over two periods of one week, one in June 1999 and one in December 1999, in primary care and emergency departments of five university hospitals and five general hospitals throughout France. Out of a total of 1,937 patients consulting, 1,562 were taking at least one drug during the previous week and were included for analysis according to the protocol. Altogether, 328 (21%; 95% confidence interval 19% to 23%) of these patients receiving at least one drug consulted because of an ADE. The sex ratio (M/F) was the same in both groups with or without ADE (1.04 vs 1.02, respectively, P = 0.83). Patients with ADE were older than those without (63.5 vs 54.8 years, P < 0.0001). Furthermore, ADR patients were more likely to have a higher severity score than no-ADE group (P = 0.019). The outcome seemed to be worse in patients with an ADE. The percentage of patients treated with 2 or more drugs and the number of drug exposures were significantly higher in patients with ADE than in those without (90.9% vs 75.0%, P < 0.0001, and 5.17 vs 3.82, P < 0.0001, respectively). The most frequent causes of visits for ADE-patients were digestive (n = 53, 16.2%), neurological (n = 52, 15.9%), cardiovascular (n = 49, 14.9%) and malaise (n = 49, 14.9%) events. In total, 410 drugs were incriminated in the occurrence of 328 ADE. The most frequently incriminated drug classes were (1) psychotropic agents, including anxiolytics and/or hypnotics, antidepressants and antipsychotics (n = 84, 20.5%), (2) diuretics (n = 48, 11.7%), (3) anticoagulants (n = 38, 9.3%), (4) other cardiovascular drugs (n = 63, 15.4%), and (5) analgesics, including non steroidal antiinflammatory agents (n = 57, 13.9%). The avoidability of ADE could be estimated by an external expert panel in 280 of the 328 cases. In 106 cases (37.9%), ADE was considered to be preventable because a contra-indication or a warning about drug use had not been respected.