Early results after rectal cancer surgery in a defined population were compared before and after the introduction of total mesorectal excision. In the first period (1984-1986) 211 cases of rectal cancer were diagnosed and in the second (1990-1992) 230. Of these, 134 patients in the first period (group 1) had anterior resection or abdominoperineal excision which was considered curative. In the second period 128 curative anterior resections and abdominoperineal excisions were performed by a limited number of surgeons familiar with total mesorectal excision (group 2). No differences between the groups were found in stage distribution, rate of curative operations, postoperative complications or postoperative mortality. Local recurrence had developed in 19 patients in group 1 and in eight in group 2, 1 year after the end of the study periods (P = 0.03). Local radicality was in doubt in 13 patients in group 1 and in eight in group 2. In the remaining 121 and 120 patients, 13 and four local recurrences respectively were present (P = 0.03). Actuarial analysis showed a significant reduction in local recurrence rate (P = 0.03) and an increase in crude survival (P = 0.03) at 4 years in group 2 compared with group 1. Total mesorectal excision reduces the local recurrence rate after excision of rectal cancer.