Cryptorchism is one of the few well-described risk factors for testicular cancer. It has been suggested that both conditions are related to increased in utero estrogen exposure. The evidence supporting the "estrogen hypothesis" has been inconsistent, however. An alternative hypothesis suggests that higher in utero androgen exposure may protect against the development of cryptorchism and testicular cancer. In order to examine both hypotheses, we studied maternal hormone levels in two populations at diverse risks of testicular cancer; Black Americans (low-risk) and White Americans (high-risk). The study population of 200 mothers of cryptorchid sons and 200 mothers of noncryptorchid sons was nested within the Collaborative Perinatal Project, a cohort study of pregnant women and their children. Third trimester serum levels of estradiol (total, free, bioavailable), estriol, testosterone (total, free, bioavailable), sex hormone-binding globulin, alpha-fetoprotein, and the ratios of estradiols to testosterones were compared between the case and control mothers. The results found no significant differences in the levels of testosterone (total, free, bioavailable), alpha-fetoprotein, sex hormone-binding globulin, or in the ratios of estrogens to androgens. Total estradiol, however, was significantly lower in the cases versus the controls (P = 0.03) among all mothers and, separately, among White mothers (P = 0.05). Similarly, estriol was significantly lower among all cases (P = 0.05) and among White cases (P = 0.05). These results do not support either the estrogen or the androgen hypothesis. Rather, lower estrogens in case mothers may indicate that a placental defect increases the risk of cryptorchism and, possibly, testicular cancer.