AbstractWing dimorphism appears in general to be determined either by a single locus, 2 allele system in which brachyptery is dominant, or by the additive action of numerous loci. In the latter case studies indicate that the heritability is typically quite large. It is generally postulated that wing dimorphism is under strong selection: why then is genetic variation not eroded? In this paper I consider three possible explanations. First, genetic variation may not be exposed to selection because environmental heterogeneity effectively makes heritability zero. Because wing dimorphisms are known to evolve it seems unlikely that this is the primary factor. Second, directional selection on a threshold trait may push the population almost to monomorphism but erodes genetic variance at a very slow rate. This mechanism cannot preserve variation but makes it possible for other factors to more easily maintain variability. Finally, I demonstrate that in a heterogeneous environment spatio‐temporal variation in fitness will itself maintain a genetic polymorphism for wing dimorphism.