OBJECTIVE: To describe a computer based system for auditing patients with colorectal cancer, to give results obtained from 1977-89, and to point out some of its advantages in terms of record keeping, evaluation of results and education of junior staff. DESIGN: Open study. SETTING: District hospital. SUBJECTS: All patients who presented with colorectal cancer. INTERVENTIONS: An initial form was filled in when the patient was discharged from hospital or died, and follow up forms were completed at each visit. MAIN OUTCOME MEASURES: Morbidity, mortality, and completion of follow up. RESULTS: 803 patients underwent 799 primary operations for colorectal cancer, of which 589 (73.3%) were potentially curative. Operative mortality was 7% (n = 56), and there were 45 leaks after 648 anastomoses (6.9%). The five year cure rate after potentially curative resections were for carcinoma of the rectum 98/196 (50%) and for carcinoma of the colon loo/212 (47%). Of 273 emergency admissions 76 required immediate operation; a quarter died postoperatively and 2 (3%) developed anastomotic leaks. There was no consistent correlation between Dukes' stage and the development of recurrences or metastases after potentially curative resections of carcinoma of the rectum, but numbers were small. CONCLUSIONS: Much valuable information can be gained from such a database. It takes a lot of extra work, but we feel that it is justified in terms of improved record keeping, evaluation of results, and education of junior staff.