The object of this paper is to suggest that there may be an unexpected connexion between parasites and the evolution of sex, using for illustration an unfamiliar type of parasite, the selfish chromosome. The major intellectual challenge of sexuality is to an environment which is continually getting worse. The elegant solution given by the Red Queen theory (Levin, 1975; Hamilton, 1980; Bell, 1982; Bell & Maynard Smith, 1988) is that the relevant aspect of the environment is provided by antagonists—pathogens, predators and competitors—which, because they can respond adaptively so as to negate any improvement that has been made, provide a constant stimulus for continued evolution. Sexuality and recombination are favoured because some of the new combinations of genes which they create are resistant to the current population of antagonists. In other respects, sex and recombination are probably highly disadvantageous: outcrossed sex is expensive because it halves the rate of transmission of genes, while recombination breaks up successful combinations of genes. It is only in certain circumstances that the necessity for continual counter-adaptation will overcome these disadvantages: in particular, the damage (reduction in fitness) caused by an antagonist must be substantial, and the amount of damage must depend on a genetic interaction between the antagonistic species. These requirements are often satisfied by host—parasite systems, where both the ecological and genetic interactions between the antagonists may be very severe and highly specific (see reviews by Day, 1974 and Burdon, 1987). It is possible, therefore, that sex and recombination are maintained in natural populations largely through the dynamics of the coevolution of hosts and their parasites. This is certainly compatible with the major ecological patterns shown by sexual systems, with outcrossed sex being more common in the sea than in freshwater, more common at low than at high latitudes, and generally more common in stable, complex, climax environments where interactions between species are expected to be more frequent and intense (Bell, 1982). However, there is as yet no evidence which conclusively supports a direct causal link between the incidence of parasitism and the rate of recombination. In particular, it has never been demonstrated that a particular parasite has the effect of eliciting, directly or indirectly, a greater rate of genetic recombination in its host. We suggest that such a parasite exists; both the parasite and its effects are well known, but have never been interpreted in the context of the evolution of recombination through host—parasite coevolution. It is in many respects a rather unusual parasite. We shall argue that B-chromosomes represent highly evolved parasitic DNA, transmitted through the germ line and often eliciting greater rates of recombination in the host genome.