From january 1986 to december 1992, 71 patients underwent direct colo-anal anastomosis as described by Parks (CAA) after total rectal resection for carcinoma: 49 men and 22 women with a mean age of 64 years (range 37-82). In 67 cases, the indication was for adenocarcinoma of the mid and low rectum, and in 4 cases for carcinoma of the upper rectum associated with a low rectal benign tumour (6 Dukes A, 36 Dukes B, 21 Dukes C, 8 Dukes D). A diverting colostomy was constructed in all cases. One patient died from pulmonary embolism (mortality: 1.4%). Anastomotic leakage occurred in 6 cases (8.5%). None of these cases required reoperation and all colostomies have been closed. Local recurrence occurred in 12 cases (17%) 6 to 34 months after CAA, of whom 4 were treated by abdominoperineal resection. Eleven patients died from local recurrence (3 cases) or distant metastasis (8 cases). Actuarial survival at 1, 2, 3, 4 and 5 years was 92%, 88%, 78%, 75% and 69% respectively. From the functional point of view, one patient underwent abdomino-perineal resection for incontinence 3 years after CAA. All the other patients were fully continent, with a mean stool frequency of 2 per day, and good gas-stool discrimination. Twenty per cent of patients presented soiling, 20% with stool frequency, and 12% with urgency. Long term functional and oncological results make CAA a good alternative to abdomino-perineal resection for mid and low rectal carcinoma.