Joint moments are commonly used to characterize gait. Factors like height and weight influence these moments. This study determined which of two commonly used normalization methods, body mass or body weight times height, most reduced the effects of height and weight on peak hip, knee, and ankle external moments during walking. The effectiveness of each normalization method in reducing gender differences was then tested. Gait data from 158 normal subjects were analyzed using unnormalized values, body mass normalized values, and body weight times height normalized values. Without normalization, height or weight accounted for 7-82% of the variance in all 10 peak components of the moments. With normalization, height and weight accounted for at most 6% of the variance with the exception of the hip adduction moment normalized by body weight times height and the ankle dorsiflexion moment normalized by body mass. For the hip adduction moment normalized by body weight times height, height still accounted for 13% of the variance (p<0.001) and for the ankle dorsiflexion moment normalized by body mass, 22% of the variance (p<0.001). After normalization, significant differences between males and females remained for only two out of 10 moments with the body weight times height method compared to six out of 10 moments with the body mass method. When compared to the unnormalized data, both normalization methods were highly effective in reducing height and weight differences. Even for the two cases where one normalization method was less effective than the other (hip adduction-body weight times height; ankle dorsiflexion-body mass) the normalization process reduced the variance ascribed to height or weight by 48% and 63%, respectively, as compared to the unnormalized data.