Relative viabilities of individuals homozygous or randomly heterozygous for wild O chromosomes derived from a marginal (Norwegian) and a central (Greek) population of D. subobscura were obtained by means of a newly prepared marker strain. In the central and marginal populations 20.8 and 28.8 percent of all chromosomes proved lethal or semilethal in homozygous condition. Mean viability was higher for +/+ random heterozygotes than for +/+ homozygotes. This remained the case for the marginal, but not for the central populations, after exclusion of the detrimental chromosomes from the calculations. The variances of viabilities were higher for homozygotes than for heterozygotes, but the test crosses with chromosomes from the marginal population had generally higher variances than those with chromosomes from the central population. No correlation was found in either populations between the action of a chromosome in homozygous condition with its action in heterozygous condition. This is interpreted as complete recessiveness of genetic load. The results are discussed in terms of the observed reduction of the inversion polymorphism which is not paralleled by a reduction in enzyme and, as shown here, by reduction in viability variation. It is thought that the heterotic effect of inversions is due to their homeostatic action, which depends less on structural genes than on higher orders of organization due to gene interaction or regulation. Whatever the causes, it is very likely that marginal populations differ from central populations with respect to their genetic system.