The large bowel is one of the major sites of cancer incidence and mortality in most western countries. In early studies most effort was expended in trying to find the causal factor, but the direction changed when Burkitt and Walker promoted the idea that the cancer could be prevented by dietary fibre (DF). Early studies to try to confirm this hypothesis were hampered by problems concerning the definition of DF, and the lack of good analytical methods to quantitate its intake. There was, however, general agreement that DF is a complex carbohydrate of plant origin that escapes small bowel digestion and so reaches the colon. It was assumed that the major plant polysaccharide, starch, is completely digested in the small bowel, but this is not true. Thus, the identity of DF as non-starch polysaccharide is not true, and other evidence suggests that it accounts for only about 25% of the true intake of DF. Assay of 'resistant starch' is fraught with difficulty and so it is better at this stage to use fibre-rich food rather than inaccurate assays of DF in epidemiological studies. In this review, I have re-examined the epidemiological literature and have found that, whereas there could be dispute over the strength of the protection given by DF, there is no doubt about the protection afforded by cereal fibre. A further analysis of data from Europe, North America and Australasia shows that the fibre-rich foods, cereals are strongly protective, as are vegetables; fruits are neutral, while starchy root vegetables, if anything, promote colorectal cancer.