BACKGROUND The consequences of leakage from low colorectal or coloanal anastomoses are reduced by the use of a loop stoma to divert the faecal stream. Controversy continues as to whether loop ileostomy (LI) or loop transverse colostomy (LTC) is the optimal method of defunctioning such anastomoses. METHODS Patients requiring defunctioning following anterior resection and total mesorectal excision were randomized to receive either LI or LTC. Comparison was made between the groups regarding the difficulty of stoma formation and closure, the recovery after stoma closure and stoma-related complications. The minimum follow-up after stoma closure was 6 months (median 36 months). RESULTS Between October 1995 and August 1999, 70 patients were randomized (LTC 36, LI 34) of whom 63 underwent stoma closure (LTC 31, LI 32). There were no significant differences in the difficulty of formation or closure, or in the postoperative recovery between the groups. However, there were ten complications related directly to the stoma in the LTC group: faecal fistula (one patient), prolapse (two), parastomal hernia (two) and incisional hernia during follow-up (five). None of these complications occurred in the LI group. CONCLUSIONS In this randomized study, the frequency of herniation before or after colostomy closure supports the choice of LI as a method of defunctioning a low anastomosis. Both methods appear to provide satisfactory protection for the low anastomosis.