During early development, testosterone plays an important role in sexual differentiation of the mammalian brain and has enduring influences on behavior. Testosterone exerts these influences at times when the testes are active, as evidenced by higher concentrations of testosterone in developing male than in developing female animals. This article critically reviews the available evidence regarding influences of testosterone on human gender-related development. In humans, testosterone is elevated in males from about weeks 8 to 24 of gestation and then again during early postnatal development. Individuals exposed to atypical concentrations of testosterone or other androgenic hormones prenatally, for example, because of genetic conditions or because their mothers were prescribed hormones during pregnancy, have been consistently found to show increased male-typical juvenile play behavior, alterations in sexual orientation and gender identity (the sense of self as male or female), and increased tendencies to engage in physically aggressive behavior. Studies of other behavioral outcomes following dramatic androgen abnormality prenatally are either too small in their numbers or too inconsistent in their results, to provide similarly conclusive evidence. Studies relating normal variability in testosterone prenatally to subsequent gender-related behavior have produced largely inconsistent results or have yet to be independently replicated. For studies of prenatal exposures in typically developing individuals, testosterone has been measured in single samples of maternal blood or amniotic fluid. These techniques may not be sufficiently powerful to consistently detect influences of testosterone on behavior, particularly in the relatively small samples that have generally been studied. The postnatal surge in testosterone in male infants, sometimes called mini-puberty, may provide a more accessible opportunity for measuring early androgen exposure during typical development. This approach has recently begun to be used, with some promising results relating testosterone during the first few months of postnatal life to later gender-typical play behavior. In replicating and extending these findings, it may be important to assess testosterone when it is maximal (months 1 to 2 postnatal) and to take advantage of the increased reliability afforded by repeated sampling.