In an isolated Drosophila population the frequency of chromosomes carrying the second chromosome inversion In(2L)t is about 60%. Three isozyme loci are nearly monomorphic in the inversion-carrying chromosomes but are highly polymorphic in other chromosomes. From known recombination frequencies and the slight polymorphism, it is estimated that the inversion-carrying chromosomes are descended from one or a few identical chromosomes introduced about 1000 generations previously. This is long enough for lethal and mildly deleterious mutants to reach equilibrium frequencies but not long enough for very weakly selected mutants. Because the difference in viability between homozygous and heterozygous chromosomes is the same, it is inferred that there is at most very weak selection for isozyme loci. This method is sensitive enough to detect selective differences of 0.0005 per locus or less. One of the most important issues in population and evolutionary genetics is the mechanism whereby protein polymorphisms are maintained in populations. At present, most population geneticists agree in that, if selection is operating in the maintenance of protein polymorphisms, it must be quite weak (1). Because the experimental resolving power is limited, the minimal selection coefficient thus far estimated is of the order of 10-2 per locus (1). We have been able to increase the sensitivity to the order of 10-4.Drosophila melanogaster populations typically carry several cosmopolitan polymorphic inversions. These are supposed to have occurred uniquely some years ago and to have migrated to various places in the world. By utilizing the history of one type of inversion in one population, it may be possible to estimate the average selection intensity for very weakly selected polymorphic loci.Briefly, the argument is this. A chromosome carrying an inversion, In(2L)t, was introduced into the population about 1000 generations ago, as determined by the known rate of crossing over and the present frequency of a gene that has been introduced into the inversion by such crossovers. This is long enough for mildly deleterious and lethal mutations to reach equilibrium frequencies (2), and indeed the distribution of homozygous viabilities in the inverted chromosome is indistinguishable from that of the standard chromosomes. On the other hand, the time is not nearly long enough for polymorphic equilibria to be established (3), and this is confirmed by the near-monomorphism of three isozyme loci in or very near the inverted region. If such loci are maintained by strong heterotic selection there should be a substantial difference in viability between homozygotes and heterozygotes for the standard, equilibrium chromosomes but not within the nearly monomorphic chromosomes with the inversion. However, no such difference was found; the viability difference between homozygotes and heterozygotes was essentially the same in both groups, within about 1%. With an estimated 350 polymorphic loci and 20% heterozygotes (4, 5), this leads to the conclusion that...