A prospective study of 7079 people aged 45-74 recruited through general practices in South Wales, Herefordshire and Edinburgh, Scotland was undertaken to test the hypothesis that faecal bile acids are implicated in the causation of large bowel cancer. The population was recruited between 1974 and 1980 and the response rate for stool collection was 67%. Bile acid analyses were performed on those cases that presented by 1990. It was decided in advance to examine the hypothesis separately for left- and right-sided bowel cancer because of known epidemiological differences between the two sites and to exclude the cases presenting within 2 years of the stool sample from the analyses because the cancer could have been present at recruitment and might have possibly affected faecal bile acid concentrations. Each case (n = 51 left-sided and 8 right-sided) was matched with three controls by age (within 5 years), sex, place of residence and time of providing the stool sample (within 3 months). Statistical analyses using conditional logistic regression showed no significant differences between the left-sided cases and controls for any of the concentrations of individual bile acids, total bile acid concentrations, faecal neutral steroids, percentage bacterial conversion and the ratio of lithocholic acid to deoxycholic acid concentrations. There was a statistically significant (P = 0.021) association of the presence of chenodeoxycholic acid (5/8 samples) in the right-sided cases compared with the controls (3/23), odds ratio 6.26 (95% confidence interval 1.19, 32.84). A high proportion of primary bile acids has also been found in other studies of patients with a genetic predisposition to proximal bowel cancer, however this pattern may also occur in low risk groups, such as Indian vegetarians, suggesting that they may predispose to right-sided bowel cancer only in the presence of other, as yet unknown factors. If bile acids are involved in the causation of large bowel cancer, they may be part of a complex set of interacting factors.