It has been widely suggested that elevated androgen levels may be critically involved in the genesis of prostate cancer. Despite the dependency of the normal prostate and of most prostatic cancers upon androgens and the fact that tumors can be produced in some rodent models by androgen administration, I will argue that, contrary to prevalent opinion, declining rather than high levels of androgens probably contribute more to human prostate carcinogenesis and that androgen supplementation would probably lower the incidence of the disease. I will also consider the possibility that the growth of androgen-independent prostate cancers might be reduced by the administration of androgens.