The cause of human colo-rectal cancer is unknown, although international and racial comparisons suggest that diet may be important. Within populations, risk of cancer is also affected by genetic factors which remain to be elucidated. Dietary fibre and NSP consumption is not always high in populations at low risk of colo-rectal cancer, but rates are fast increasing with westernization (and meat and fat consumption) in Japan. The suggestion that dietary fibre is protective in colo-rectal cancer is based on the fact that cereal fibre from bran increases faecal weight, dilutes large intestinal contents, and speeds up transit time. In animal models, bran reduces the number of tumours induced by chemical carcinogens, and cellulose may have a similar effect. The faeces of some individuals contain mutagens, some of which have been identified as fecapentaenes and heterocyclic amines. Bran reduces faecal mutagenicity, although the mutagen concerned is unknown. Most dietary fibre is fermented in the large gut by anaerobic bacteria and little remains in faecal matter. Recent observations have shown that substantial amounts of starch survive digestion in the small bowel and are available also for fermentation in the large gut. The metabolic consequences of fermentation may be important in carcinogenesis via altered N metabolism, SCFA production, and pH reduction. Methane is also produced in some individuals, but, contrary to previous findings, is not a risk factor for large bowel cancer. Starch appears to be beneficial as a substrate for fermentation because yields of the SCFA butyrate are increased both in vitro and in vivo. Butyrate is an energy substrate for the colonic mucosa and an anti-proliferative and differentiating agent in cell culture lines. Possible mechanisms whereby starch and NSP may protect against colo-rectal cancer, therefore, exist. The majority of individual case-control epidemiological studies suggest that fibre-containing foods are protective in colo-rectal cancer, although this effect is largely due to vegetable, rather than cereal, consumption. Case-control studies of diet and large bowel cancer may, however, reflect the effect rather than the cause of the disease, so that confirmation of the possible protective effects of starch and NSP is needed from accurate prospective studies both of diet and associated risk factors.