Forty-five patients with obstructing carcinoma of the colon and rectum were compared with 176 patients with nonobstructing tumor diagnosed over a 10-year period. The age and sex distribution did not differ between the two groups. The site of greatest risk for obstruction was the splenic flexure and descending colon (50%); median survival in these patients was one-half that for other sites. In the obstructed group no patient had Dukes' stage A vs. 9.6% in the nonobstructed patients, and 22% of the obstructed patients had Dukes' D vs. 14% of the nonobstructed patients. The 5-year survival in Dukes' A was 82%, while no survivors were found for Dukes' D. The crude 5-year survival rate was 22.4% in the obstructed patients and 49.1% in the nonobstructed patients; in-hospital mortality was 22.4% and 6.8%, respectively, and the adjusted actuarial survival was 39% vs. 64%, respectively. Curative resection was performed in 68% of the obstructed and 83% of the nonobstructed patients. The adjusted actuarial 5-year survival rates for these patients were 53% and 76%, respectively. The criteria for the tumor grade tested in this study, which included differentiation of the tumor cells, size, and the presence of perforation, did not influence the survival. Twenty-one patients underwent primary resection, 15 had stage resection, and 9 had diversion procedures. The in-hospital mortality rates were 35% for primary resection and diversion procedures and 7% for stage resection. The crude 5-year survival was 32% for primary resection, 42% for stage resection (not statistically significant), and 0 for diversion procedures. The poor prognosis for the obstructed patients in our study was mainly related to 1) high in-hospital mortality, 2) the lower rate of curative resection, 3) unequal distribution of the tumor site, and 4) to a small extent, the difference in Dukes' stage.