Epidemiologic evidence on the relation between nutrition and colorectal cancer is reviewed. Colon cancer varies approximately 20-fold internationally. Although there is clear evidence of genetic predisposition to colon cancer, much of this variation appears to be related to differences in dietary habits. At present, the data suggest that vegetables are associated with lower risk, and that fiber alone does not account for this association. Further, meat consumption is associated with increased risk but this, too, is not explained solely by its fat content. Several microconstituents of the diet may be associated with reduced risk--including folate and calcium--but phytochemicals of other sorts may be relevant. Mutagenic compounds, particularly heterocyclic amines, produced when protein is cooked, plausibly explain the meat association. The most consistent inverse association is with physical activity. Alcohol is associated, though inconsistently, with increased risk. Rectal cancer is less well studied but, at present, there are few data to suggest that the dietary risk factors are markedly different. Physical activity does not appear to be associated with a lower risk. Colorectal adenomatous polyps also appear to share the spectrum of risk factors seen with colon cancer, although, for adenomas, tobacco smoking is also a clear and consistent risk factor. There are a variety of links between the dietary epidemiology and physiology of colorectal neoplasia and the relevant pathologic and molecular changes. Other causal connections remain to be explicated.