The Jackson-Pollock-Ward (JPW) generalized equations for men and women (1978, 1980) were developed primarily with lean, non-Hispanic white (NHW) subjects. Our research (2008) confirmed that the JPW equations give biased percent fat (BF%) estimates when applied to ethnically diverse young men and women with higher levels of obesity. PURPOSE: To develop valid prediction equations to estimate BF% measured with dual-energy X-ray absorptiometry (DXA) in diverse, contemporary young adults. METHODS: The subjects included 435 men and 712 women from the TIGER study who ranged in age from 17 to 35 y. The racial/ethnic composition for men and women, respectively, was: NHW, 43.5% and 32.9%; Hispanic, 28.1% and 30.8%; and African-American (AA), 39.0% and 25.7%. The subjects BF% was measured from 1 to 3 times over 9 months (total observations = 701 for men and 1,144 for women). Linear mixed models (LMM) regression was used to develop two different equations using DXA BF% as the dependent variable. Model 1 included sum of skinfolds (SS; chest, abdomen, and thigh for men; triceps, suprailium and thigh for women). Model 2 included BMI and waist (WC, men) or hip circumference (HC, women). Race/ethnic group was included as a categorical variable in each model with NHW as the referent. RESULTS: For men, both the SS and BMI equations resulted in a regression weight for AA men that was statistically significant compared to NHW men, while the regression weight for Hispanic men was within chance variation of NHW men. The standard errors of estimate were 3.08% for the SS model and 3.98% for the BMI model in men. For women, DXA BF% of Hispanic women was 2.5% higher than NHW for a given SS, while the regression weight for AA women was within chance variation of NHW women. The relation between BMI and BF% was quadratic, and hip circumference was related to BF% independent of BMI. Controlling for BMI and HC, the BF% of AA women was 1.7% lower and the Hispanic BF% was 1.7% higher compared to NHW women. The standard errors of estimate were 4.0% for the SS model and 3.8% for the BMI model in women. CONCLUSIONS: The LMM results confirmed that race/ethnicity accounted for significant BF% variance beyond that accounted for by SS or the combination of BMI and WC or HC. The developed SS and BMI equations provide accurate models to estimate DXA BF% in ethnically diverse young men and women.