A total of 175 patients who underwent a curative resection for a colonic (n = 130) or a rectal cancer (n = 45) between 1986 and 1992 were entered into a routine colonoscopy program. Colonoscopies were performed 1 year after the operation, and then at 2-year intervals. The findings at colonoscopy, as well as those of preoperative colonoscopy (when performed), were recorded. Eleven anastomotic recurrences were diagnosed at an asymptomatic stage, at a mean follow-up of 14 months. All of them were identified in patients with a stage B or C primary rectosigmoid cancer. Eight patients underwent another potentially curative re-operation. Only perioperative colonoscopy (preoperative colonoscopy; first postoperative colonoscopy in patients for whom the preoperative procedure was incomplete or not performed) allowed diagnosis of second cancers (n = 7) and adenomatous polyps greater than 10 mm (n = 17). Further colonoscopies detected only polyps less than 10 mm. Positive examination rates for successive follow-up colonoscopies were 15, 20 and 23%, respectively; they were significantly higher in patients who had previously had adenomatous polyps than in patients who had not: 30% versus 6% (P < 0.025), 46% vs 5% (P < 0.005) and 38% vs 11% (P < 0.025), respectively. From these data, the following recommendations are made: (1) All colorectal cancer patients should have a total colonoscopy either before (whenever possible) or soon after operation; (2) Based on results of the perioperative colonoscopy, patients: should undergo their first follow-up colonoscopy only 3 yearly (presence of synchronous adenomatous polyps) or 5 yearly (absence of synchronous adenomatous polyps) after resection; (3) In patients with stage B or C primary rectosigmoid cancer, a surveillance of the suture line by rigid proctosigmoidoscopy should be added during the first 2 postoperative years: 6, 15 and 24 months after the operation.