Cox regression analysis was used in the study of 151 cases of surgically treated colorrectal carcinoma. A "curative" resection was performed in all of them. Clinical data from medical records, histology, biochemical determinations, and intraoperative findings were assessed. Four of the studied variables were significantly related to tumor recurrence: Dukes stage, tumor spread, histological variety and preoperative CEA level. Age, sex, tumor location, surgical technique, lymphocytes in the peripheral blood, preoperative levels of acute phase reactants and hepatic enzymes did not provide information about the final outcome of these patients. An index of prognosis was statistically obtained in order to identify a high risk of recurrence group of patients, who may benefit from adjuvant therapy.