Previous studies of human populations have failed to find a significant relationship between genetic variability, as measured by total heterozygosity, and cistron size, as measured by subunit molecular weight of proteins, but the number of different rare alleles in human populations has been shown to be correlated with subunit size. The present paper examines these relationships further, utilizing data on electrophoretic variants at 27 loci for 12 human populations with a total of 800 000 individual system observation. The results indicate that, if genetic variability is measured by rare allele heterozygosity instead of total heterozygosity, there is a significant correlation with subunit size. In addition, there are significant differences for rare allele heterozygosity between multimeric and monomeric proteins, the range of variability being less in the multimers (and in the total) than for monomers. Finally, rare allele heterozygosity has a much bigger range of variability than the range of subunit size. By contrast, the range of rare allele heterozygosity between populations is less than ten-fold, a factor not evident in effective population sizes. Both interlocus and interpopulational estimates of relative electromorph mutation rates (REMR) have been calculated, utilizing the distributions of the number of different rare alleles as well as rare allele heterozygosity. The range of these estimates are much lower than the estimates given by Zouros (1979) using total heterozygosity as input.