We developed a method to estimate genotype-specific average relative mortality risk, R, from genotype distributions in cross-sectional studies of people belonging to different age-groups, and applied the method to new data from a study of apolipoprotein E genotypes (apoE) in 177 Danish centenarians and data from a study of 40-year-old Danish men. Twenty-one percent of the centenarians were epsilon 2-carriers (genotypes epsilon 2 epsilon 2 and epsilon 3 epsilon 2) and 15% were epsilon 4-carriers (genotypes epsilon 4 epsilon 4 and epsilon 4 epsilon 3) compared to 13 and 29%, respectively, of the young men. The R-values were 0.95 (95% CI 0.88 to 1.02) for epsilon 2-carriers and 1.13 (95% CI 1.05 to 1.22) for epsilon 4-carriers, using epsilon 3 epsilon 3- and epsilon 4 epsilon 2 genotypes as reference. Corresponding values for epsilon 4-carriers were obtained by using published data from a French and a Finnish study of centenarians, whereas the values for epsilon 2-carriers were about 0.90 with these data. The method to estimate mortality risk and the results associate with the view that the apoE gene is a "frailty gene." On the other hand, if odds ratios are used to summarize data from studies of this kind, they are more impressive and may propagate the misconception that apoE is a "longevity gene".