The problem of extrapolating effects of reproductive toxins on experimental animals to predict the doses that would produce infertility in human males is discussed using published data on effects of testosterone and estradiol on sperm production in the rat, rabbit, rhesus monkey, ram, stallion, and humans. This analysis indicates that calculation of the dose of testosterone that reduces human sperm counts by a given percentage is best done using the dose administered to laboratory animals expressed on the basis of body weight, as opposed to some other parameter such as body surface area. A survey of the available data in the literature indicates the incompleteness of the data set and the specific information needed to improve the basis for extrapolation. Nevertheless, we can predict from studies on laboratory animals the dose of testosterone necessary to reduce sperm counts in humans within a factor of 2.