We have used the aberrant crypt focus (ACF) assay to test and develop hypotheses linking diet and colon cancer. The hypotheses were suggested by epidemiological studies that identified possible dietary factors associated with colorectal cancer risk. The ACF assay was used to quantitate the effect of the dietary factors on the initiation and growth of these putative precursors of colon cancers in experimental animals. Using this approach we have developed 3 new hypotheses for the role of diet in colorectal cancer. These are (1) a risk associated with 5-hydroxymethyl-2-furaldehyde in caramelized sugar, (2) a risk associated with some factor in thermolyzed casein, and (3) a risk associated with single nutrient boluses of sucrose and fructose. The importance of these hypotheses has still to be tested in long term carcinogenesis experiments, in analytic epidemiology studies and then, perhaps, in intervention trials.