Recent publications dealing with sexually transmitted HIV consider how the average number of sexual partners, as well as the rate of partner exchange, can determine the evolution of increased mutation rates, genetic (and antigenic) diversity, and consequently, increased virulence of the virus (Ewald 1991, 1994). It is intuitive that in a population of relatively monogamous individuals HIV would benefit from remaining latent. If such outside sexual activity takes too long to occur, only the pathogen variants that have some way to extend their potential infectiousness would be transmitted venereally. When sexual activity increases so that people are involved with different partners more frequently, the factors contributing to virulence shift. A particular strain, or variant, of a sexually transmitted virus, like HIV, that actively reproduces in an infected person could soon infect several new hosts, whereas a strain that is latent for several years would propagate less extensively, if at all. Therefore, increased number of sexual partners and/or rates of partner exchange would favor HIV variants that replicate rapidly. Here fast replicating, highly cythopatic strains that are not recognized by the immune system have a disproportionately larger effect (Nowak et al. 1990). This evolutionary process leads eventually to the development of AIDS after a variable incubation period. So, the pressures favoring the evolution of fast viral reproduction when sexual contact is high and of delayed viral reproduction when sexual contact is low provide the basis for a prediction: HIV virulence should be correlated with rates of sexual contact (Ewald 1994). Actually, literature concerning the capacity of HIV variants to induce AIDS (and eventually kill their hosts) with different rates show that AIDS progression is associated with the history of sexual intercourse (Perkins et al. 1987; Ward et al. 1989; van Griensven et al. 1990; Blower and Boe 1993; Ewald 1994). Obviously, the rationale exposed above is also valid for parenteral transmission of HIV. As the number of people sharing injecting equipment, as well as the rate of exchanges increase, the fast replicator strains of HIV are favored. This is an example of how a host behavior can act as a selective pressure for increased virulence of some parasites. Here I provide a mathematical treatment to Ewald's (1994) verbal arguments for the evolution of virulence of HIV. It assumes that an evolutionary mechanism-on a very fast time scale-is responsible for viral pathogenesis (Nowak et al. 1990). In the first analysis, I propose an adaptation of Levin and Pimentel's (1981) model for the study of the evolution of virulence of HIV. More recently, Nowak and May (1994) analyzed the same model for the study of superinfection and the evolution of parasite virulence. In the second analysis, this model is rewritten in terms of Volterra-Lotka interspecific competition equations, as done by Massad et al. (1993) for the study of the evolution of resistance against antibiotics. Both analyses yield the same result: strains which are transmitted by more active hosts (either sexually or parenterally), evolve towards a higher virulence.