Developmental synapse elimination in the rat levator ani (LA) muscle is sensitive to gonadal androgen. This process occurs faster in castrated male rats that lack gonadal testosterone and is largely prevented by testosterone treatment. Because testosterone can be irreversibly converted into either androgenic metabolites such as dihydrotestosterone or estrogenic metabolites such as estradiol, the present experiment sought to determine which of these metabolites account for testosterone's effect. Male rat pups at postnatal day 7 (P7) were castrated and given daily subcutaneous injections of one of 5 possible treatments for 3 weeks (P7-P28): (1) testosterone propionate (TP), (2) dihydrotestosterone propionate (DHTP), (3) estradiol benzoate (EB), (4) a combination of DHTP and EB or (5) sesame oil vehicle. At the end of treatment, the LA and extensor digitorum longus (EDL) muscles were dissected and their motor nerve terminals were stained with tetranitroblue tetrazolium. Hormone effects on synapse elimination were evaluated by counting the number of motor axons that contacted individual muscle fibers. The lumbosacral spinal cord was also dissected and processed histologically to examine the motoneurons in the spinal nucleus of the bulbocavernosus (SNB), which innervates the LA. Hormone effects on SNB motoneuron size were assessed by measuring the cross-sectional area of SNB motoneuronal somata and nuclei. We report that DHTP mimics the effects of TP on synapse elimination in the LA muscle, but that EB, acting either alone or together with DHTP, has little or no effect on this process. Synapse elimination in the EDL was unaffected by any hormone treatment. TP or DHTP, but not EB, increase the size of SNB motoneurons. We conclude that testosterone or its androgenic metabolites influence synapse elimination in the LA and probably exert these effects via androgen receptors.