In order to determine the incidence of infection following sclerotherapy and the role of antimicrobial prophylaxis, a prospective randomized control study was performed comparing i.v. imipenem/cilastatin, with an infusion of dextrose-saline as a control group. One hundred patients with bleeding esophageal varices were included. All episodes of infection were documented during admission to the unit. Ninety-seven patients were evaluable. Post-sclerotherapy bacteremia developed in six (5.6%) of 107 sclerotherapy sessions in the control group and one (1.1%) of the 88 sclerotherapy sessions in the imipenem/cilastatin group (P < or = 0.1, NS): six of these seven post-sclerotherapy bacteremias occurred after emergency sclerotherapy. Infection within 7 days of the procedure was documented after 43 (22.1%) of the 195 sclerotherapy sessions, 18 (20.5%) in the imipenem/cilastatin group and 25 (23.4%) in the control group (P = NS). These infections were significantly more common after emergency sclerotherapy, 40 (34.8%) of 115 sessions, than after elective sclerotherapy, three (3.8%) of 80 sessions (P < or = 0.0001). A short prophylactic antibiotic regime does not reduce the risk of early bacteremia or the frequency of infection after sclerotherapy. The higher risk of infection after emergency sclerotherapy may be therefore related more to the gastrointestinal hemorrhage and its associated effects than to sclerotherapy.