There are a number of examples of genetic polymorphism in which a balance seems to be maintained, at least in part, by frequency-dependent selection. The selective value of each gentoype is then negatively related to its frequency. Mimetic polymorphisms are perhaps the best known examples (Sheppard, 1958), but there is evidence that frequency-dependent selection acts upon other polymorphic systems. It seems that parasites and predators may attack disproportionately many of a common morph, and disproportionately few of a rare one (Haldane, 1954; Clarke, 1962a, b). This has been termed apostatic selection (Clarke, 1962a). Clarke and O'Donald (1964) have set up some simple mathematical models of balanced polymorphisms under frequency-dependent selection. In this paper I will suggest a more general model, and consider the effects of such selection upon the dominance of polymorphic genes. In a polymorphism which involves mimicry, if there are no disturbing influences, there will be a theoretical equilibrium frequency for each mimetic form. I shall call this the focal frequency. The values of these foci will depend upon the relative numbers and palatability of the models as well as the effectiveness of the mimics. In an apostatic polymorphism there will also be foci whose values will depend upon the susceptibility of the different forms to the selective agent (e.g., their relative palatability and crypsis). In both situations we can regard the foci as the phenotype frequencies at which the frequency-dependent selective values of the different forms become equal (see below). It must be emphasized that the attainment of stable polymorphism under frequency-dependent selection does not necessarily imply that an "adaptive peak" (in any sense of the term) has been reached. Wright (1942, 1949) has shown that such selection may lead to a value of W which is less than the possible maximum. Models of frequency-dependent selection. -The most general model to be considered [section (iii)] shows the effects of frequency-dependent selection upon alleles with incomplete dominance or intermediate heterozygotes. Using it, we can discover the effects of such selection upon the dominance of polymorphic genes. Before doing so, however, it will be useful to consider two simpler models which are special cases of model (iii). The first of these deals with the situation (probably uncommon) in which the heterozygote is phenotypically distinct from both homozygotes, resembles neither of them, and tends towards an independent selective focus. In the second model, one allele is assumed to be completely dominant to the other. (i) If the heterozygotes are subject to an independent selective focus, the effects of mimetic or apostatic selection can be represented as follows: