An excess rate of mediastinitis complicating cardiac operations occurred in one of two hospitals served by the same surgeons. The rates were 4% at hospital A and 0.48% at hospital B for the same period (P = .002). At hospital A five patients who underwent a thoracic reoperation experienced postoperative pneumonia or bacteremia prior to onset of mediastinitis; these infections were caused by the same bacterial species. This circumstance provided a unique opportunity for comparing infection control practices, in which one hospital served as a control for interventional changes. At hospital A regular and temporary critical care nursing personnel, who took care of both infected and uninfected patients, did not antisepticize their hands before manipulation of percutaneous catheters. After cohorts of nurses were formed and antisepsis of the hands with alcohol was strictly enforced (no diminution in the number of temporary personnel was instituted), the frequency of mediastinitis decreased significantly among patients who underwent reoperations at hospital A (P = .002), but no concurrent change was noted at hospital B. Preventable postoperative remote-site infection may lead to mediastinitis.