The effects of superoxide dismutase on aging were tested using two different experimental approaches. In the first, replicated populations with postponed aging were compared with their controls for frequencies of electrophoretic alleles at the SOD locus. Populations with postponed aging had consistently greater frequencies of the allele coding for more active SOD protein. This allele was not part of a segregating inversion polymorphism. The second experimental approach was the extraction of SOD alleles from different natural populations followed by the construction of different SOD genotypes on hybrid genetic backgrounds. This procedure did not uncover any statistical effect of SOD genotype on longevity or fecundity. There were large effects on longevity and fecundity due to the family from which a particular SOD genotype was derived. To detect the effects of SOD genotypes on longevity with high probability would require a ten-fold increase in the number of families used.