We examine the assumption of "dominance" with regard to viability of the Cy and Pm marker chromosomes in D. melanogaster . This assumption is often invoked for the extraction of wild-type second chromosomes from natural populations and for the calculation of relative viability indices. Significant genotypic variances for viability are found among both Cy/+j and Pm/+i heterozygotes in California and Japanese populations. The magnitude of the Pm/+ i genotypic variance is substantially less than that of the Cy/+j heterozygotes (less than one half). Significant reciprocal effects are also found to influence Cy/+j, Pm/+i and +i/+j viabilities. We conclude that viability indices of heterozygotes based on the Curly method are biased. We suggest that viability indices in the future be expressed relative to the viability of the Cy/Pm genotype (Curly-Plum method) or possibly that of the Pm/+i genotype (Plum method).