The antibiotic management of 139 consecutive patients with presumed viral meningitis evaluated during a 6-month period was examined. The presupmitive diagnosis of viral meningitis was made in retrospect by consensus among the authors, using clinical and routinely available laboratory informations. Sixty-eight (51.9%) of 131 patients with complete records were treated with antibiotics after diagnosis, 25 for 2 days or less and 43 for longer than 2 days. Antibiotic treatment was retrospecitvely judged to be unjustified in 35 (81.4%) of the 43 patients treated for longer than 2 days. When compared with untreated patients antibiotic treatment was started in younger female children with lower cerebrospinal fluid glucose values and longer duration of symptoms. There was no difference between the two groups in other cerebrospinal fluid values, peripheral white blood cell count or history of preceding antibiotics. In contrast no associations were found with treatment beyond 2 days, compared with treatment for 2 days or less. Thus the decision to stop antibiotic treatment early did not appear to be made according to consistent clinical ceriteria. This apparent lack of consistent criteria suggests the need to develop clinical guidelines for such decisions, both to aid clinicians and to provide standards for medical audit.