Epidemiological and clinical studies support the view that aggressive acts like suicidal and violent behaviors share a common substrate. Certain aspects of violence in males have been related to high testosterone levels, but the relation of testosterone to attempted suicide has not been studied until now. We estimated plasma levels of testosterone (T), LH, and FSH in 80 male subjects after a suicide attempt and in whom a psychiatric assessment was done during their hospitalization. Suicide intent was evaluated in 72 subjects. A group of 56 healthy males in the same age range served as control. As a group, attempters showed significantly lower T levels, marginally higher LH, and normal FSH compared to controls. The attempters who used violent methods (26 subjects) had T levels even lower than the non-violent (drug overdose) subgroup. Comparisons of T levels of subgroups according to the (main) drug ingested (analgesics, benzodiazepines, antidepressants, neuroleptics, or other drugs) did not reveal any significant drug effect. In relation to diagnosis, the lowest T levels were found in the subgroup with schizophrenia (29 subjects). The T levels of this subgroup were also significantly lower compared to those of a group of 31 male schizophrenic patients, hospitalized and treated with neuroleptics. If the influence of post-attempt stress and medical condition on plasma T could be ruled out, low plasma T may prove to be a biological predictor of attempt, at least in male schizophrenic patients. Nevertheless, the findings differentiate suicidal behavior from other aggressive/violent behaviors and do not support the notion that suicidal and aggressive behaviors are manifestations of the same impulse.