According to conventional wisdom, parasites and pathogens should evolve reduced virulence to their hosts, because more virulent parasites and pathogens are more likely to drive their hosts, and themselves, to extinction. But this view has been criticized for its reliance on group selection. According to an alternative perspective, selection will favor whatever level of virulence maximizes the rate of increase of the parasite or pathogen. This optimum virulence depends on the functional relationship between a parasite or pathogen's transmissibility and its effect on host mortality, with selection often favoring an intermediate degree of virulence. The thesis of this paper is that models in which intermediate levels of virulence are favored lead quite naturally to the further conclusion that parasites and pathogens should-up to a point-become less virulent over time, once the feedbacks between ecological and evolutionary processes are incorporated into the analysis. As a consequence of successive adaptations by the parasite or pathogen, the density of susceptible hosts is reduced, thereby altering the balance between selective forces so as to favor reduced virulence. However, the evolutionarily stable strategy that is achieved is bounded away from complete avirulence. We conclude that models in which intermediate virulence is favored do not necessarily contradict the conventional wisdom in the long run; in fact, these models provide a simple mechanistic explanation for the evolution of reduced virulence.