Gene variants found to associate with human longevity in one population rarely replicate in other populations. The lack of consistent findings may partly be explained by genetic heterogeneity among long-lived individuals due to cohort differences in survival probability. In most high-income countries the probability of reaching e.g. 100years increases by 50-100% per decade, i.e. there is far less selection in more recent cohorts. Here we investigate the cohort specificity of variants in the APOE and FOXO3A genes by comparing the frequencies of the APOE ε4 allele and the minor alleles of two variants in FOXO3A at age 95+ and 100+ in 2712 individuals from the genetically homogeneous Danish birth cohorts 1895-96, 1905, 1910-11, and 1915. Generally, we find a decrease in the allele frequencies of the investigated APOE and FOXO3A variants in individuals from more recent birth cohorts. Assuming a recessive model, this negative trend is significant in 95+ year old individuals homozygous for the APOE ε4 allele (P=0.026) or for the FOXO3A rs7762395 minor allele (P=0.048). For the APOE ε4 allele, the significance is further strengthened when restricting to women (P=0.006). Supportive, but non-significant, trends are found for two of the three tested variants in individuals older than 100years. Altogether, this indicates that cohort differences in selection pressure on survival to the highest ages are reflected in the prevalence of longevity gene variants. Although the effect seems to be moderate, our findings could have an impact on genetic studies of human longevity.