The purpose of the study was to examine the effect of prophylactic penicillin treatment on the infection rate in patients with traumatic wounds of hands or feet with underlying lesions of bone, tendon or joint. A total of 599 patients were randomized to double-blind treatment with either (A) one injection of penicillin G, (B) penicillin V tablets for 6 days or (C) no antibiotic treatment. In group A the infection rate was 4.9 per cent, in group B 6.7 per cent and in group C 10.2 per cent. The difference between the infection rates in group A and C was 5.3 per cent with a 95 per cent confidence interval (CI) from 0.1 to 10.4 per cent (P = 0.046). Correction for an interim analysis and subgroup analyses increases this P value to 0.096. The difference between the infection rates in group B and C was 3.6 per cent (95 per cent CI from -1.9 to 8.9, P = 0.20). Patients in group B had gastrointestinal complaints five times as often as patients in group A and C. Prophylactic tablets for 6 days require patients' co-operation while an injection ensures that the prophylaxis is given. We therefore conclude that a single prophylactic injection of two million units of penicillin G is advisable for patients with these injuries.