The subjects of this study were 58 healthy boys, 15-17 years-old, selected from the public school districts of Solma, a suburb of Stockholm, to provide a roughly random sample of the total male student population of the 9th grade. The subjects provided two sets of blood samples for plasma testosterone assays. About 1 month before the blood samples were drawn, the subjects completed a number of personality inventories, and highly reliable peer ratings of habitual aggressive behavior and physical strength were available. The main findings of the study were the positive associations of testosterone level with the self-report scales of verbal and physical aggression, mainly reflecting responsiveness to provocation and threat, and lack of frustration tolerance. Lower positive correlations were obtained with peer ratings of aggressive behavior and a self-report scale of antisocial behavior. These findings suggest that testosterone levels have a direct impact on verbal and physical aggression provoked by some threat, while they indirectly affect unprovoked aggression and antisocial behavior through the mediating variable of lack of frustration tolerance. The findings are compared with similar studies. Tabular data and 17 references are provided.