Models of sexual selection typically assume a cost to the production and maintenance of sexually selected traits. However, within a population, individuals of high phenotypic quality may not only show greater expression of the trait but may also have higher survival rates. In such a case, the true cost of the trait may be difficult to detect. Comparative studies can circumvent some of the difficulties inherent in intraspecific studies. In this study, I used a comparative approach to examine the relation between sexual size dimorphism and adult mortality in natural populations of mammals. After controlling for the effects of phylogeny, I found that the degree of male-bias in adult mortality was positively correlated with the degree of sexual size dimorphism across taxa. This is the first comparative evidence for a cost of sexually selected traits in natural populations of mammals. In contrast to the general rule that mortality in mammals is male-biased, I also found that it is frequently female-biased in monogamous taxa of mammals. I suggest that in monogamous species the cost of reproduction to females will be similar to that of polygynous species. However, males of monogamous species will be spared the costs of male-male competition and dimorphic growth. This decrease in male costs relative to females may then lead to a female-bias in mortality.