OBJECTIVE To compare three doses of cefotaxime alone with a single dose of cefotaxime and metronidazole for the prophylaxis of infection after elective colorectal operations. METHODS Prospective random control trial. METHODS Hillerød and Frederiksberg Hospitals, Copenhagen, Denmark. METHODS 660 consecutive patients who were to undergo elective colorectal operations during a 48 month period (January 1987-January 1991); 93 (14%) were withdrawn after randomisation leaving 567 for assessment. METHODS Mechanical bowel preparation, and then either cefotaxime (Claforan) 2 g intravenously at induction of anaestesia and 3 and 9 hours later (n = 280) or a single dose of cefotaxime 2 g plus metronidazole (Flagyl) 1.5 g intravenously at induction of anaestesia (n = 287). RESULTS 44 patients in the cefotaxime group developed wound infection (16%) compared with 19 (7%) in the combined group (p < 0.001). In the cefotaxime group 22 of the 241 patients who had an anastomosis developed leaks (9%) compared with 8 of the 239 in the cefotaxime/metronidazole group (3%). There were no differences in the incidence of intra-abdominal abscesses or burst abdomens. The most common organisms isolated from wounds were Escherichia coli and Bacteriodes fragilis. CONCLUSIONS One dose of cefotaxime and metronidazole is active against a wide range of organisms and resulted in significantly fewer wound infections than three doses of cefotaxime alone.