In this study, 417 patients undergoing "clean" elective neurosurgical operative procedures were randomized to receive a broad-spectrum antibiotic (piperacillin) or placebo given as three perioperative doses, each 6 hours apart. Randomization was carried out by hospital pharmacists, and the investigators remained blinded until the end of the study. Twenty cases were excluded from analysis because either an unforeseen second operation was performed or antibiotic therapy was initiated within 30 days after surgery to treat infection or the risk of infection. Twelve of the 205 patients treated with placebo developed postoperative wound sepsis, and four of the 192 piperacillin-treated patients developed wound sepsis--a statistically significant difference (p less than 0.05, Fisher's exact test). Piperacillin thus appeared to reduce the incidence of neurosurgical wound infection in this study.