The hypothesis of this study was that the position of rectal cancer within the circumference of the rectum influences mortality. Tumor position was prospectively documented in 181 patients with rectal carcinoma by two examiners. The results were analyzed for correlation to survival using the Lifetest model and for multivariate correlation using the Cox regression model. An anterior tumor location was present in 43 patients and was found to have a significantly higher survival rate than other positions. Two-thirds of anterior tumors were of pathologically favorable Dukes' stages. Fifty-five patients had posterior tumors with decreased survival rates, two-thirds of which were of unfavorable stages. Circumferential position in 61 patients was most predictive of poorer outcome, with a relative risk of death being increased by 4.6 times (P = 0.014) and a 5-year survival rate of 68.8 per cent; 85 per cent of these tumors were of pathologically unfavorable stages. The 5-year survival rate for the whole group, which included 181 patients with all histopathological stages except those with distant metastases at presentation, was 78.5 per cent. This ranking of survival rates was found to be consistent in each category with the postoperatively determined Dukes' stage, which carried a prognostic significance of P = 0.0001. We conclude that tumor position is a significant indicator of prognosis available before surgery for rectal cancer.