Scholars have noted the need to examine the psychometric properties of measures that can be used in evaluating moral education programs. The present study was designed to examine the best-fitting factor model of a commonly-used measure of prosocial moral reasoning (PROM) across samples from Brazil and the USA, gender and adolescent age groups. The samples consisted of 619 college students (M age = 20.59 years, SD = 4.08; 41% men, 59% women) and 239 middle and high school students (M age = 14.02 years, SD = 3.04; 45% boys, 55% girls) from the USA. There were 114 college students (M age = 21.81, SD = 4.33; 35% men, 65% women) and 136 middle and high school students (M age = 14.93 years, SD = 1.55; 42% boys, 58% girls) from Brazil. A series of (multigroup) confirmatory factor analyses were conducted to test the best fitting factor structure of the PROM and the invariance of this factor structure across culture, gender and age groups. Evidence for measurement invariance was found such that a four-factor model was a slightly better fitting model than the five-factor model across all groups. Discussion focuses on theoretical and methodological implications of the findings. (Contains 2 tables.)