Studies of migrants to Australia indicate that the risk of colorectal cancer (CRC) can be influenced by environment during adult life. Under a gene x environment interaction model for the risk of CRC, it is postulated that a high proportion of the variation in CRC incidence among populations is attributable to differences in diet, and that the average genetic susceptibility of populations to dietary carcinogenic components is similar. The possible relevance within populations of individual susceptibility to dietary components is supported by studies showing that the risk of CRC, compared with that of the general population, is not increased in the spouses of cases. It is postulated that, conditional upon a relevant dietary exposure, the level of susceptibility to diet is a major determinant of variation in individual risk. Despite high correlations of meat and fat with CRC among populations, the results of case-control studies have been inconsistent. Inability to stratify subjects by susceptibility may explain the lack of association of intake of fat and meat with CRC in many studies. A new generation of case-control and cohort studies of diet and CRC may extend the limits of epidemiology. Prevention trials with beta-carotene and other compounds have not protected against colorectal neoplasia, and may have exerted adverse effects. This possibility challenges the philosophy of chemoprevention trials in favor of trials of changes in diet towards patterns that are associated with lower risk.