The traditional method for calculating risk in prospective and retrospective studies is based on the assumption that the study population is homogeneous. Risk is therefore estimated as an overall average for the entire population, when in fact some individuals may be at high risk and others at little or no risk. This paper introduces an alternate approach to risk estimation. The calculations are equally simple and utilize the same data. Yet, the new approach allows for heterogeneity and can detect it when it exists. The new method was applied to HIV seroconversion data from a follow-up study, age-at-onset distribution for Huntington disease, and age-specific prevalence of insulin-treated diabetes. These analyses were intended to demonstrate both applicability of the method to different types of data and the accuracy of the estimates when compared with the known parameters. The HIV analysis predicted a high-risk subgroup constituting about 17% of the cohort. This estimate closely approximates the actual 16% who reportedly engaged in high-risk activities and had a 15-fold higher seroconversion rate than the rest of the cohort. There is no evidence from genetic linkage studies for heterogeneity in Huntington disease. The present results, however, suggested that 14%-18% of individuals who are susceptible to the disease have a much lower risk than others. Diabetes data was chosen because the model is clearly too simplistic for this disease, and the analysis did reveal lack of fit of the model.