This paper studies the design of the optimal non linear taxation in an economy where longevity varies across agents, and depends on three factors: longevity genes, health investment and farsightedness. Provided earnings, farsightedness and genes are correlated, governmental intervention can be justi.ed on two grounds: correction for a lack of farsightedness and redistribution across both earnings and genetic dimensions. Whether longevity-enhancing spending should be subsidized or taxed is shown to depend on the combined effects of myopia, self-selection and free-riding on the annuity returns. Our policy conclusions depend also on how productivity and genes are correlated, on the complementarity of genes and efforts in the survival function,...