Multiplicative Poisson models were used to identify subgroups of the Danish population at high and low risk of developing cancer of the right or left side of the colon, and cancer of the rectum. The analysis was based on cross-linked data from the 1970 census and the Danish Cancer Registry, where a 10-year follow-up period yielded some 20,000 colo-rectal cancers, in approximately 2.5 million persons. The risk of cancer of the right side of the colon in longer educated men living in apartment houses was almost twice as high as in farmers living in single family houses (relative risk 1.84; 95% confidence interval 1.42-2.37). A two-fold ratio (RR 2.18; 95% CI 1.70-2.62) was also seen in the risk of cancer of the left side of the colon between men with longer education in Greater Copenhagen and farmers in Jutland. The annual number of colon cancer cases in men in Denmark could be reduced by 27% if the incidence for all men was equal to that found for the low risk group of farmers.