&NA;Between 1978 and 1981, 73 patients with colonic or rectal cancer were randomized to have their anastomoses made by either a single interrupted layer of braided polyester sutures, or by a circular stapling instrument. Of these operations, 20 were considered to have been palliative, the remaining 53 being potentially curative. The incidence of local recurrence in the latter group was analyzed in relation to initial septic and anastomotic complications. The 53 patients were followed for a median of 36 months (range, 1 to 87); 24 were alive and well and 22 had died of disseminated cancer or unrelated causes. Seven patients died with local recurrent disease proved at laparotomy or autopsy after a median of 33 months (range, 3 to 72). Thirty anastomoses were stapled and 23 sutured; of the seven patients who died with local recurrent disease, six had stapled anastomoses (Fisher's exact probability F2=0.12; log rank chi‐square=3.53, 0.05<P<0.10). Two patients who died with locally recurrent disease had had clinically apparent anastomotic leads and one other patient had had a radiologically demonstrated leak. This compares with a total of seven leaks (clinical or radiologic) in the remaining group of 46 patients with no recurrence (Fisher's exact probability F2=0.11). These results tend to support the hypothesis that anastomotic leaks may lead to locally recurrent disease, particularly after stapled anastomoses