To determine the effect of the ventilation system on infection rates after total hip and total knee arthroplasties performed in operating rooms with and without a horizontal unidirectional filtered air-flow system, using modern antiseptic conditions and antibiotic prophylaxis, all of the single-stage procedures (3175 of a total of 4769) were subjected to statistical analysis and fifty-seven matched pairs for controls were established. A reduced infection rate after total hip replacement (from 1.4 to 0.9 per cent) and an increased infection rate after total knee replacement (from 1.4 to 3.9 per cent) were found when patients operated on in the filtered laminar air-flow operating room were compared with those whose operations were done in two conventional rooms. This pattern was statistically significant and was believed to be due to the positions of the operating team and of the wound with respect to the air flow. Prospectively accumulated factors (such as the experience of the surgeon, the duration of surgery, the diagnosis, and the patient's age) as well as retrospectively accumulated factors (such as predisposing conditions of the patient) did not explain the observed patterns of infection.