A prospective study of wound infections following major head and neck cancer surgery was undertaken to define a rational approach to trials of antibiotic prophylaxis and initial therapy of these infections. Preoperative aerobic cultures were taken from the planned site of skin incision and from the oropharynx. Both aerobic and anaerobic cultures were obtained from all wound infections. Patients receiving prophylactic antibiotics were excluded from the study. The data indicate that preoperative cultures are not usually predictive of the bacteriology of subsequent wound infection. Mixed aerobic and anaerobic flora were cultured from most wound infections and usually reflected normal anaerobic oropharyngeal flora and/or exogenously acquired Staphylococcus aureus. Bacteroides fragilis was not cultured in this series. Antibiotics selected for trials of prophylaxis or initial treatment of these infections should cover both the resident oral aerobic and anaerobic flora and S aureus. Coverage for B fragilis does not appear necessary. Antibiotic choices might include (1) penicillin G agents plus a penicillinase-resistant penicillin or (2) a parenteral cephalosporin.