Background: Colorectal cancer is a common cause of cancer death in Australia and is primarily managed operatively. Surgical databases are valuable in monitoring performance in cancer treatment and detecting problems and trends. Methods: Diagnostic and treatment variables and short-term outcomes were gathered prospectively for patients undergoing resection for colorectal cancer over a 9-year period. Survival data were obtained by linkage to state and interstate death indices. Results: Eight hundred and five patients underwent resection for colorectal cancer during the study period. Overall 5-year survival was 61%. Five-year cancer-specific survival was 73%. Five-year cancer-specific survival for Australian Clinico-Pathological Staging (ACPS) stages A, B, C and D was 96, 80, 61 and 19%, respectively (P < 0.0001). Emergency presentations showed diminished survival (59% versus 75%, P < 0.0001) after controlling for age and stage (hazard ratio (HR) 1.78, P= 0.005), as did transfusion recipients (63% versus 74%, P= 0.0014; HR 1.78, P= 0.004). Anastomotic leak did not affect survival in multivariable analysis. Non-cancer causes accounted for 26% deaths, primarily comprising cardiovascular deaths in the elderly. Discussion: High case ascertainment, data completeness and accuracy can be obtained with prospective, independently gathered data linked electronically to national death records. Survival for colorectal cancer in South Australia continues to improve. Close follow-up for disease recurrence is warranted for transfusion recipients, emergencies and advanced disease. Locally managed databases with linkage to state registries and other institutions are powerful methods to improve data quality and surgical care at a national level.