Summary We propose and evaluate the hypothesis that the CAG repeat region of the androgen receptor represents a locus of antagonistic pleiotropy in the context of sexual selection and sexual conflict. Short repeats are associated with increased transactivation of the androgen receptor at the molecular level, and increased fertility at the phenotypic level. However, short repeats are also associated with increased risk of prostate cancer, and with more aggressive forms of the disease. The somatic evolution of cancer cell lineages also shows a repeated pattern of shortening of the CAG repeat in association with cancer progression, apparently as a result of positive selection among cell lineages. We further postulate that other genes associated with prostate cancer are likely to mediate antagonistic pleiotropy in the context of sexual selection and sexual conflict. A key prediction of this hypothesis is that the genes mediating antagonistic pleiotropy will show historical evidence of positive selection, particularly in the context of sexual conflict. Previous research on the molecular evolution of specific genes associated with prostate cancer supports this prediction, and we suggest further critical tests of the role for genomic conflicts and tradeoffs in the evolution of cancer risk.