The wealth of new information about the molecular mechanisms of carcinogenesis that has been obtained over the last decade has generated a variety of new hypotheses about the relationship between diet and cancer. The classical model for the anticarcinogenic effect of cereal fibre is based largely on assumptions about the physical effects of nonstarch polysaccharides within the colonic lumen, and on the putative benefits of faecal bulk. However, the difficulty of studying the disease process directly in human subjects continues to limit our ability to test these hypotheses directly. The role of mutagenesis due to oxygen free radicals in the pathogenesis of human cancers remains largely hypothetical, and attempts to prevent cancer by intervention with high doses of antioxidant vitamins have been largely unsuccessful. Various promising lines of evidence reviewed briefly here do suggest that the presence of substances other than nonstarch polysaccharides in cereal cell walls may at least contribute to the epidemiological relationship between cereal fibre consumption and cancer. Among the various topics reviewed, the intraluminal production of biologically active phenolic compounds and the antioxidant and antiproliferative effects of phytic acid seem particularly worthy of further study in the context of high-fibre cereals.