The association of perioperative blood transfusion and cancer recurrence was analyzed in a 10-year series of 520 patients with colorectal carcinoma undergoing curative surgery. Cancer recurrence was observed in 35 of 165 (19%) non-transfused and in 119 of 355 (34%) transfused patients (p < 0.001). A significant difference in favor of the non-transfused group was also evident in cases of colonic (p < 0.05) and rectal (p < 0.01) tumors separately. the transfused group had both more recurrences and higher mortality. Although there were no differences in age, sex, and Dukes's stage between the two groups, further analysis showed other significant variables explaining the different result. First, there were significantly more patients with rectal cancer in the transfused group. in addition, emergency operations for obstruction and resections of adjacent organs for tumor invasion were significantly commoner in transfused patients. Exclusion of such patients and those with small tumors excised locally per anum resulted in the disappearance of the prognostic difference between the transfused and non-transfused groups (marginally significant difference in rectal carcinoma; p = 0.05). It is concluded that the seemingly adverse effect of blood transfusion on the prognosis of colorectal carcinoma may largely be explained by other associated variables.