Antibiotics were given for prophylaxis in hip-fracture patients to a randomized group of patients with hip fractures (417) treated by operation at the Massachusetts General Hospital during three years. After excluding those patients with history of penicillin sensitivity, and those patients already receiving antibiotics at the time of fracture, 348 patients were randomly assigned to a treatment (sodium nafcillin) or control (glucose) group. Either antibiotic or placebo was given before, during, and after operation. The incidence of postoperative wound infection was seven in 145 (4.8 per cent) in the control group, and one in 135 (0.8 per cent) in the treatment group, a statistically significant difference (P = 0.041). The incidence of postoperative wound hematomas was high in each group, possibly because of the routine use of sodium warfarin. Subsequent infection developed in a few of these hematomas, with similar incidence in the control (four in 145) and treatment (six in 135) groups.