Cigarette smoking and alcohol consumption both associate, at the individual level, with esophageal cancer; the basis of these associations is explored. Secular (temporal) trends in England and Wales, from 1911 to 1980, show that death rates from esophageal cancer correlate positively with alcohol consumption but negatively with average rates of cigarette smoking. The details suggest that alcohol helps to cause esophageal cancer through an indirect precipitating action. This hypothesis is tested, and loosely corroborated, by using the sex- and age-patterns of esophageal cancer in England and Wales and in U.S. Whites to infer the sex- and age-patterns of alcohol consumption by the victims of the disease. Genetic predisposition, alcohol consumption around 2-3 years before death and a proximal precipitator all appear to have a causal role.