Improvement of health care policy requires an assessment of health care practices. In France, morbidity registries might be the best tool for such an assessment. This study shows how the treatment of rectal cancer can be assessed by French cancer registries. Two studies were conducted: a cross-sectional study on data from 7 cancer registries in 1990 and a longitudinal study on data from 2 digestive cancer registries (departments of Calvados and Côte d'Or) between 1978 and 1990. Finally, we conducted a regional audit concerning quality control in rectal resection for cancer in Lower Normandy between 1988 and 1993. In 1990 the mean resection rate was 77.8%. The sphincter preservation rate was also significantly increased to 53.9% in 1990. The use of adjuvant radiotherapy significantly increased between 1978 and 1990, more rapidly in university centres. In more recent years, the use of radiotherapy concerned 50% of resected rectal cancers with no differences between the various types of health care centres. However, in 1990, major geographical variations were observed for the use of adjuvant radiotherapy. Similar geographical variations were observed for the use of chemotherapy which did not increase with time. Rectal cancers were not diagnosed earlier from 1978 to 1990 in the two departments of Calvados and Côte d'Or. The use of reproducible quality criteria (length of distal excision, number of nodes examined and histological status of lateral margins) showed a global deficiency and marked variations between the various types of health care centres and levels of surgical training. The french network of French cancer registries (FRANCIM) provides accurate and reliable knowledge on medical practices, geographical variations, trends and quality control. The potential of cancer registries has not been clearly determined, although such information is required to plan health care policy.