OBJECTIVE Although visceral adipose tissue (VAT) is the component of body composition most highly associated with cardiovascular risk factors, its measurement requires expensive procedures, such as magnetic resonance imaging. This study examined the ability of simple demographic and anthropometric measurements to predict magnetic resonance imaging-derived VAT in 76 apparently healthy, black and white youths with obesity who were 7 years to 16 years of age. RESEARCH METHODS AND PROCEDURES Stepwise multiple linear regression was used to develop a prediction equation for VAT based on 13 simple anthropometric variables (height, weight, body mass index, triceps skinfold, calf skinfold, sagittal diameter, waist circumference, hip circumference, thigh circumference, waist/hip ratio, waist/thigh ratio, sagittal diameter/thigh ratio, and percent body fat from the sum of calf and triceps skinfolds) and three demographic variables (age, gender and ethnicity). RESULTS The stepwise multiple regression procedure yielded a final model that included two anthropometric variables (sagittal diameter and waist/hip ratio) and one demographic variable (ethnicity). The prediction equation was: VAT = - 124.06+ 16.67 (ethnicity)+4.15 (sagittal diameter)+100.89 (waist/hip ratio), where ethnicity was coded as 0= black and 1 = white. The model explained 63% of the variance in VAT and was associated with a measurement error of 23.9%. DISCUSSION Although the model seems to lack sufficient explanatory power for routine use in clinical settings with individual patients, it may have some utility in epidemiological studies given its relatively small (<25%) standard error of estimate.