The purpose of this study was to compare as anti‐infectious prophylaxis in elective colorectal cancer surgery the effect of metronidazole alone and in combination with ampicillin, and the effect of a duration of 1 or 3 days of prophylaxis. The prophylactic regimens designated regimens A‐D given in randomized order were metronidazole 500 mg used alone or with 2.0 g ampicillin administered every 8 hours as separate but simultaneous infusions. All patients studied received preoperative mechanical evacuation of bowel contents. Eight surgical departments participated in the study. Two hundred thirty‐three patients were studied. The distribution of sex, age, and type of operation was similar among the groups of patients receiving each regimen, except that there were more cases of sigmoidectomy, low anterior resection, or rectal amputation in the group receiving regimen D. The duration of the operations was comparable, even for each type of operation considered separately. Samples for bacteriological examination were obtained by abscess punctures when relevant. The pus was taken and transported to the laboratory under anaerobic conditions. Moderate or severe infections were observed in 6 (10.3%) of 58 patients on regimen A, in 2 (3.5%) of 58 patients on regimen B, in 4 (7.0%) of 57 receiving regimen C, and in 2 (3.3%) of 60 given regimen D. The highest incidence of postoperative infections was observed in rectal amputation. The bacteria causing postoperative infections were similar in the regimens A and C receiving only metronidazole for 1 and 3 days, respectively and in regimens B and D in which ampicillin was added. Only one anaerobe, aClostridium perfringens, was recovered from regimen C; twenty‐two strains of anaerobic bacteria were recovered from regimen A. The number of aerobic bacteria was 25 in regimen A and 16 in regimen C. The yield of bacteria was much more sparse when metronidazole was combined with ampicillin. Eleven isolates (2 anaerobes) were recovered from regimen B, only one isolate was recovered from patients on regimen D, an indole positiveProteus. In conclusion, th'is study indicates that a combination of metronidazole and ampicillin is particularly useful in rectal surgery. Metronidazole alone may suffice in colonic surgery, but a combination with an agent against aerobes is recommended in rectal surgery. The difference between 1‐day prophylaxis and 3‐day prophylaxis was insignificant for metronidazole plus ampicillin; a single day of this prophylactic regimen would appear advisable.