Conventional models predict that low genetic relatedness among parasites that coinfect the same host leads to the evolution of high parasite virulence. Such models assume adaptive responses to hard selection only. We show that if soft selection is allowed to operate, low relatedness leads instead to the evolution of low virulence. With both hard and soft selection, low relatedness increases the conflict among coinfecting parasites. Although parasites can only respond to hard selection by evolving higher virulence and overexploiting their host, they can respond to soft selection by evolving other adaptations, such as interference, that prevent overexploitation. Because interference can entail a cost, the host may actually be underexploited, and virulence will decrease as a result of soft selection. Our analysis also shows that responses to soft selection can have a much stronger effect than responses to hard selection. After hard selection has raised virulence to a level that is an evolutionarily stable strategy, the population, as expected, cannot be invaded by more virulent phenotypes that respond only to hard selection. The population remains susceptible to invasion by a less virulent phenotype that responds to soft selection, however. Thus, hard and soft selection are not just alternatives. Rather, soft selection is expected to prevail and often thwart the evolution of virulence in parasites. We review evidence from several parasite systems and find support for soft selection. Most of the examples involve interference mechanisms that indirectly prevent the evolution of higher virulence. We recognize that hard selection for virulence is more difficult to document, but we take our results to suggest that a kin selection model with soft selection may have general applicability.