The EEA-autostapling device was used for the anastomosis in 34 patients operated upon with anterior resection for carcinoma of the rectum. The lower border of the tumours was located between 6 and 20 cm from the anal verge and the distal margin free of tumour in the resected specimen was 4 cm (1-8 cm). The anastomoses were within 10 cm from the anal verge in 27 patients. Technical problems occurred in two patients. Eight cases showed incomplete tissue rings but anastomotic leakage was noted merely in four patients (12%). Wound infections were rare (6%). According to Dukes' classification the material comprised 15% Dukes' A, 32% Dukes' B. 38% Dukes' C and 15% Dukes' D. The patients were followed with clinical examinations, rectal palpation and sigmoidoscopy every 3-6 months after the operation. Early local recurrence (median value 5 months) was diagnosed in seven patients (21%). They had in common tumour in the lower part of the rectum, extramural spread and a short distal margin free of tumour. Three of these were classified as Dukes' D, three as Dukes' C and one as Dukes' B. It is concluded that early local recurrence after anterior resection may be due to local tumour spread but a decreased circular radicality may be another reason. The results demonstrate the need to define patients who might be treated by an abdomino-perineal excision even if the EEA-stapler makes a low rectal anastomosis possible.