An empirical study investigated the association between testosterone and aggression in a sample of young male medical students: 101 volunteers completed the Aggression Questionnaire [Buss et al, (1992) Journal of Personality and Social Psychology 63:452-459] and the Conflict Tactics Scales [Straus (1979) Journal of Marriage and the Family 41:75-88] and gave a blood sample to measure testosterone levels, Neither total testosterone nor the free androgen index correlated significantly with any of the aggression subscales (physical, verbal, anger, and hostility) or with the Conflict Tactics Scales measure, This finding applied to both Caucasian and Asian respondents. Since this was one of several null results among an overall finding of a positive association, a meta-analysis of previous studies involving student samples and those from aggression-prone populations was undertaken. A weighted mean of d =.40 was obtained from 18 studies, but no differences were found between the magnitude of effect in the two types of sample. Of several other study characteristics, only the source of the testosterone assay influenced effect size magnitude, those from salivary samples being higher than those from plasma samples. (C) 1998 Wiley-Liss, Inc.