Two recent reviews reached opposing conclusions regarding the apparent detrimental effect of peri-operative blood transfusion on patient survival after resection of colorectal cancer. However, both sides agree that the influence of confounding variables has not been controlled adequately in most studies. The present study is based on 433 patients who had a curative resection for colorectal cancer between 1984 and 1989 and whose details and follow-up information were recorded in the Concord Hospital Colorectal Cancer Registry, Sydney. The association between peri-operative transfusion and patient survival after resection of colorectal cancer was examined taking into account 20 other prognostic variables. A proportional hazards regression model showed that transfusion did not have a statistically significant independent effect on survival after controlling for the patient's poor general condition, the complexity of the surgery and the occurrence of postoperative complications.