Desmoid tumors occur at a much higher frequency in association with familial adenomatous polyposis than with any other disease. However, published estimates of risk have varied widely. In this study we have examined the occurrence of desmoids in a population of polyposis patients in order to determine an accurate estimate of the risk of desmoid disease. The crude frequency of desmoids in our population was about 6%, but the risk was dependent on the age and sex of the individuals at risk. The lifetime risk of incurring a desmoid was 8% for males and 13% for females. The distribution of occurrence of desmoid disease with regard to age was unimodal in each sex in contrast to the multimodal distribution observed for desmoids in nonpolyposis patients. Females were more likely to be affected than males and at an earlier age. We also detected heterogeneity of risk between families segregating for the polyposis gene, with the desmoid disease tending to cluster in some of those families. Consequently crude estimates of risk should be used with caution.