The relationship between coffee drinking and the risk of digestive tract neoplasms was analyzed in a case-control study of 50 cases of cancer of the mouth or pharynx, 209 of the esophagus, 397 of the stomach, 455 of the colon, 295 of the rectum, 151 of the liver, 214 of the pancreas, and 1944 control subjects admitted for acute, non-digestive tract disorders. There was no significant or consistent association between coffee and cancers of the mouth or pharynx, esophagus, stomach, liver, or pancreas. In particular, for pancreatic cancer, the multivariate relative risks for the intermediate and upper tertiles were 1.05 and 1.01, respectively. There were significant inverse trends in risk with measures of coffee consumption for colon and rectal cancers, the multivariate relative risks according to tertiles of coffee consumption being 0.86 and 0.64 for colon and 0.97 and 0.66 for rectum. This apparent protection is in agreement with some (but not all) previous epidemiological evidence and finds a possible biological interpretation in terms of interference on bile secretion, causing reduced bile acid and neutral sterol concentrations in the bowel. In conclusion, the results of this study, the major interest of which lies in the opportunity of drawing up an overall pattern of risk for various digestive neoplasms, offer further reassurance as regards the effects of coffee on digestive tract carcinogenesis.