A total of 127 patients (57 given placebo and 70 given a single preoperative dose of 1 g cefazolin) undergoing thoracic surgery were included in a randomized double-blind trial. The two groups were similar in regard to mean age, sex ratio, in-hospital stay before surgery, underlying disease, risk factors, type of surgery, mean duration of surgical procedure, and mean duration of chest tube drainage. The relative risk of wound infection of the patients from the placebo group was 3.27 (range 1.5-11.5; 95% confidence interval). Cefazolin significantly reduced (p less than 0.01) the wound infection rate--1 case (1.5%) in the cefazolin group versus 8 cases (14%) in the placebo group--but not the incidence of postoperative pleural empyema--5 (7%) versus 8 cases (14%)--or nosocomial pneumonia--3 (4%) versus 5 cases (9%). Cultures were made from 3 out of 9 wound infections and Staphylococcus aureus or S. epidermidis was isolated in all 3. In addition, cultures were made from 6 out of 13 pleural cavity infections and S. aureus (1 case) or other microorganisms (5 cases) were isolated in all 6. Mortality was similar in both groups and all deaths unrelated to the infections. No adverse side effects of the drug were encountered. In conclusion, a single preoperative dose of 1 g cefazolin proved to be effective for reducing the wound infection rate in non-cardiac thoracic surgery.