In a prospective, double-blind study, 193 patients hospitalized for established hand infections were randomized to receive either cefamandole intravenously followed by cephalexin by mouth or methicillin intravenously followed by dicloxacillin by mouth. Careful aerobic and anaerobic cultures were performed. Multiple organisms grew in cultures from 84% of the patients (over three isolates per infection on average). Human bite wounds contained anaerobes 43% of the time compared with 12% for other wounds. The majority of wounds (72%) required operative treatment. In 128 patients assessable for treatment outcome, results were unsatisfactory in 11 (9%). There was no difference in outcome between cefamandole (6/59, 10%) and methicillin (5/59, 8%). The presence of anaerobes, Eikenella corrodens, human bites, or an increasing number of organisms was associated with an unsatisfactory response. The presence of Staphylococcus aureus and/or beta-hemolytic streptococci was associated with a favorable response. The incidence of antibiotic-resistant isolates did not correlate with outcome.