Low anterior resection with total mesorectal excision for rectal carcinoma is associated with a high anastomotic leakage rate, and the effectiveness of a defunctioning stoma in preventing anastomotic leakage remains controversial. In this study a policy of selective defunctioning stoma for stapled colorectal anastomosis after low anterior resection with total mesorectal excision in 148 consecutive patients was evaluated prospectively. A defunctioning stoma was performed in 61 patients (41%) considered at high risk of anastomotic leakage. Clinical leakage occurred in 2 patients (3.3%) with a stoma and 11 patients (12.6%) without a stoma (p = 0.047). Among those without a stoma, the leakage rate among male patients (20.9%) was significantly higher than that for female patients (4. 5%) (p = 0.022). Leakage subsided with conservative treatment in the two patients with a stoma, but seven patients without a stoma developed peritonitis requiring laparotomy. No deaths resulted from leakage, and there was one hospital death (0.6%) in the whole group. Median hospital stay was similar with and without a stoma (13.0 vs. 12.0 days) (p = 0.290). Closure of the stoma was associated with no mortality, a morbidity rate of 8.7%, and a median hospital stay of 6. 0 days. In conclusion, a defunctioning stoma is effective in preventing clinical anastomotic leakage after low anterior resection with total mesorectal excision. The relatively high incidence of leakage in the low risk group indicates the difficulty of predicting anastomotic leakage and hence the need for more liberal use of a defunctioning stoma especially in male patients.