Anecdotal reports of "roid rage" and violent crimes by androgenic steroid users have brought attention to the relationship between anabolic steroid use and angry outbursts. However, testosterone effects on human aggression remain controversial. Previous studies have been criticized because of the low androgen doses, lack of placebo control or blinding, and inclusion of competitive athletes and those with preexisting psychopathology. To overcome these pitfalls, we used a double-blind, placebo-controlled design, excluded competitive athletes and those with psychiatric disorders, and used 600 mg testosterone enanthate (TE)/week. Forty-three eugonadal men, 19-40 yr, were randomized to 1 of 4 groups: Group I, placebo, no exercise; Group II, TE, no exercise; Group III, placebo, exercise; Group IV, TE plus exercise. Exercise consisted of thrice weekly strength training sessions. The Multi-Dimensional Anger Inventory (MAI), which includes 5 different dimensions of anger (inward anger, outward anger, anger arousal, hostile outlook, and anger eliciting situations), and a Mood Inventory (MI), which includes items related to mood and behavior, were administered to subjects before, during, and after the 10 week intervention. The subject's significant other (spouse, live-in partner, or parent) also answered the same questions about the subject's mood and behavior (Observer Mood Inventory, OMI). No differences were observed between exercising and nonexercising and between placebo and TE treated subjects for any of the 5 subdomains of MAI. Overall there were no significant changes in MI or OMI during the treatment period in any group. Conclusion: Supraphysiological doses of testosterone, when administered to normal men in a controlled setting, do not increase angry behavior. These data do not exclude the possibility that still higher doses of multiple steroids might provoke angry behavior in men with preexisting psychopathology.