OBJECTIVE To determine the procedure of choice for rectal cancer, particularly low rectal cancer. METHODS Complete search, according to evidence-based methods, of comparative studies and national surveys published in English since 1990. METHODS comparative studies between abdominoperineal excision (APER) and sphincter-saving operations (SSO) with a minimum of 50 patients presenting cancer in the lower one-third of the rectum, perfect split of cases with cancer located in the lower, middle or upper one-thirds of the rectum, specified numbers of patients treated by surgery alone or combined with radio-chemotherapy, specified length of follow-up with a minimum of 1 year, univariate or multivariate analysis of prognostic factors. Thirty-four studies fulfilling evidence level C were analyzed, including 6,570 patients. METHODS operative risk, local disease control, disease free or cancer specific survival and quality of life. RESULTS Postoperative morbidity after APER and SSO is comparable and postoperative mortality decreased to 2% or less. The type of surgery was not identified as a prognostic factor in terms of local disease control and survival. Quality of life is significantly inferior after APER. National data reveal an APER rate for cancer of the whole rectum (up to 16 cm) at 50% or above, and SSO still would represent only 32% of the radical resections for low rectal cancer. CONCLUSIONS All available evidence indicates that SSO should be the procedure of choice for rectal cancer, even in the lower one-third. An APER should only be performed when cancer invades the anal sphincters and negative resection margins cannot be achieved by a SSO.