Between 1986 and 1988, 450 adults undergoing coronary artery bypass, cardiac valve replacement, or both were enrolled into a prospective, randomized, comparative trial of cephalothin versus cefamandole as perioperative prophylaxis. They were assessed during their hospitalization and at 6 weeks and 6 months after discharge for postoperative infectious complications. Eleven patients had major postoperative infections including 5 with sternal wound infections (three bacteremic), 6 with bacteremia, 1 with prosthetic valve endocarditis, and 3 with severe venous donor graft site infections. Eight major infections occurred in patients receiving cephalothin prophylaxis and three in patients receiving cefamandole, with all five sternal wound infections occurring in the cephalothin group. Postoperative pathogens responsible for the major infections included gram-negative aerobes in 5 patients, Staphylococcus aureus in 4, and Staphylococcus epidermidis in 2. Preoperative colonizing staphylococcal isolates were not predictive of postoperative staphylococcal pathogens. Although there was no statistically significant difference in rate of major postoperative infectious complications using either cephalothin or cefamandole prophylaxis, there was a trend in favor of cefamandole. Gram-negative aerobes are becoming increasingly important pathogens in this setting.