OBJECTIVE--To assess the differences among surgeons in postoperative complications, postoperative mortality, and survival in patients undergoing surgery for colorectal cancer. DESIGN--Prospective study of patients with colorectal cancer managed by one of 13 consultant surgeons, none of whom had a special interest in colorectal surgery. SETTING--Royal Infirmary, Glasgow. PATIENTS--645 sequential patients with colorectal cancer presenting over the six years from 1974 to 1979. MAIN OUTCOME MEASURES--Postoperative complications, postoperative mortality (within 30 days), and survival (up to 10 years); predictive factors for postoperative mortality and survival; and relative hazard rate ratios for individual surgeons. RESULTS--The proportion of patients undergoing apparently curative resection varied among surgeons from 40% to 76%; overall postoperative mortality varied from 8% to 30%. After curative resection postoperative mortality varied from 0% to 20%, local recurrence from 0% to 21%, and the rate of anastomotic leak from 0% to 25%. Survival at 10 years in patients who underwent curative resection varied from 20% to 63%, two year survival in those who underwent palliative resection varied from 7% to 32%, and median survival in those who underwent palliative diversion varied from one to eight months. The hazard rate ratios among individual surgeons, taking into account the identified risk factors, varied from 0.56 to 2.03, from 0.17 to 1.92, and from 0.57 to 1.50 for curative resection, palliative resection, and palliative diversion, respectively. CONCLUSION--There were significant variations in patient outcome among surgeons after surgery for colorectal cancer; such differences compromise survival. A considerable improvement in overall survival might be achieved if such surgery were undertaken by surgeons with a special interest in colorectal surgery or surgical oncology.