Abstract.-By means of population genetical models, we investigate the competition between sex-specific segregation distorters. Although the models are quite general, they are motivated by a specific example, the t complex of the house mouse. Some variants at this gene complex, the t haplotypes, distort Mendelian segregation in heterozygous males in their favor. The selective advantage at the gamete level is counterbalanced by strong negative fitness effects at the individual level (male sterility or even lethality in both sexes). A plethora of different t haplotypes has been found, both in the field and in the lab. Up to now, however, models have focused on the equilibrium frequency of a single t haplotype. In contrast, we explicitly model the competition between several t haplotypes. A deterministic model for a large, well-mixed population predicts a surprisingly high degree of polymorphism. Haplotypes with seemingly inferior fitness characteristics may easily coexist with "superior" haplotypes. For instance, a lethal haplotype with a low segregation ratio may stably coexist with a sterile haplotype with a high segregation ratio. Stable coexistence is even possible for haplotypes with a segregation disadvantage. A simple stochastic model shows that the same principles apply in the context of a structured metapopulation. Although counterintuitive at first sight, all our results can be explained by the fact that segregation distorters have an inherent advantage when they are rare. We conclude that fitness comparisons are not sufficient to predict the outcome of competition when selective forces are acting at different levels. The t complex of the house mouse is a large complex of closely linked genes on chromosome 17. Variants of this gene complex, the so-called t haplotypes, lead to reduced viability and complete male sterility in homozygous condition. Nevertheless, t haplotypes reach appreciable frequencies (10%-25%) in most feral house-mouse populations (Lenington et al. 1988). This polymorphism results from the fact that the negative fitness effects of t haplotypes at the individual level are counterbalanced by a fitness advantage at the gamete level. In heterozygous males, Mendelian segregation is strongly distorted in favor of t haplotypes. In fact, segregation ratios greater than 0.95 are not uncommon.By now, a large number or different t haplotypes has been found. All t haplotypes lead to male sterility in homozygous condition, but they vary in their effects on viability. Some haplotypes are completely viable, some have a reduced viability, whereas still others induce complete lethality in homozygous condition (Klein et al. 1984). Although a large variety of fitness effects has been described, most t haplotypes in feral house-mouse populations lead to embryonic lethality both in males and in females (Silver 1993).The segregation ratio is also subject to considerable variation: segregation ratios vary from as low as 0.20 to as high as 1.00 (Bennett et al. 1983;Gummere et al. 1986;Lenington and Heisler 1991...