Summary Diets high in animal fat and protein and low in fibre and calcium are thought to be factors in the etiology of colorectal cancer. Intakes of these nutrients were determined in three groups participating in a randomised trial of faecal occult blood (FOB) screening. A diet history was obtained by interview from 147 patients with colorectal adenomas, 153 age and sex matched FOB-negative controls (a) and 176 FOB-positive controls without colorectal neoplasia (b). Unconditional logistic regression was used to estimate relative risks (RR) and 95% confidence limits (t) adjusted for age, sex and social class.After adjustment for total energy intake, no associations were found with total, saturated or monounsaturated fat, or calcium intake. For total fibre intake there were non-linear relationships with both control groups with the crude RR for highest quintiles of total fibre intake compared to the lowest being 0.6, although this pattern was no longer apparent after adjustment for energy intake with group (a). In comparison with group (b) cereal fibre intake showed a more consistent inverse relationship with adenoma prevalence with the RR for ascending quintiles of intake being 1.0, 0.7 (0.3-1.6), 0.5 (0.3-1.1), 0.7 (0.4-1.4) and 0.3 (0.1-0.6) (trend x2 = 8.80, p = 0.003). In comparison with group (a), the adjusted RR for the highest quintile of cereal fibre intake compared with the lowest was 0.6, but no clear trend was apparent. There was an unexpected positive relationship between adenomas and polyunsaturated fat intake with the RR for having an adenoma being 1.0, 2.8 (1.3-6.1), 1.6 (0.7-3.4), 3.5 (1.6-7.5) and 2.3 (1.1-5.0) for ascending quintiles of polyunsaturated fat intakes (trend x2 = 4.8, P = 0.03) in comparison with group (a) only.Our data, while providing no support for the role of dietary animal fat or protein, do support the protective role of dietary cereal fibre in the etiology of colorectal adenomas.High intakes of animal fat and protein (Wynder & Shigematsu, 1967;Drasar and Irving, 1973), and low intakes of fibre (Burkitt, 1971) and calcium (Newmark et al., 1984) have been postulated to increase the risk of colorectal cancer. However, the role of these nutrients has not been clarified in analytical epidemiological studies (Zaridze, 1983;Willett, 1989a (Paul & Southgate, 1978).In addition to the nutrients about which we had specific hypotheses, we considered total energy intake as an important potential confounder (Willett, 1989b, pp , the sources of protein and fibre, and the different types of fatty acid.We also asked subjects about the frequency of consumption of certain foods which might be markers of 'healthy Correspondence: J. Little,