This study examined the extent to which factors presumed to be correlated with body mass index (BMI) vary across four race- and gender-specific groups. Data were drawn from the American Changing Lives Survey to estimate separate multivariate regression models for the total study sample that included African-American males, Caucasian males, African-American females and Caucasian females. The dependant variable of interest was BMI. Independent variables included age, human capital variables, relationship and support measures, health status and behavior measures, and stress and outlook measures. Results from the pooled model indicated that BMI was associated with a number of factors such as employment status, chronic illness, financial strain and religiosity. However, race- and gender-specific regression models revealed that predictors of BMI varied considerably for African-American men, Caucasian men, African-American women and Caucasian women. In other words, these models disentangled important correlations not observed in the pooled model. These findings suggest that addressing racial disparities in body weight-related outcomes requires health practitioners to modify obesity prevention and treatment efforts to incorporate a broader array of factors inherent to specific racial and gender populations.