CONTEXT: An increasing number of studies suggest that bone marrow adipose tissue (BMAT) might play a role in the pathogenesis of osteoporosis. Our previous study of Caucasian women demonstrated that there is an inverse relationship between BMAT and whole-body bone mineral density (BMD). It is unknown whether visceral adipose tissue (VAT), sc adipose tissue (SAT), and skeletal muscle had an effect on the relationship between BMAT and BMD. OBJECTIVE: In the present study we investigated the relationship between pelvic, hip, and lumbar spine BMAT with hip and lumbar spine BMD in the population-based Coronary Artery Risk Development in Young Adults (CARDIA) sample with adjustment for whole-body magnetic resonance imaging (MRI)-measured VAT, SAT, and skeletal muscle. DESIGN: T1-weighted MRI was acquired for 210 healthy African-American and Caucasian men and women (age 38-52 yr). Hip and lumbar spine BMD were measured by dual-energy x-ray absorptiometry. RESULTS: Pelvic, hip, and lumbar spine BMAT had negative correlations with hip and lumbar spine BMD (r = -0.399 to -0.550, P < 0.001). The inverse associations between BMAT and BMD remained strong after adjusting for demographics, weight, skeletal muscle, SAT, VAT, total adipose tissue (TAT), menopausal status, lifestyle factors, and inflammatory markers (standardized regression coefficients = -0. 296 to -0.549, P < 0.001). Among body composition measures, skeletal muscle was the strongest correlate of BMD after adjusting for BMAT (standardized regression coefficients = 0.268-0.614, P < 0.05), with little additional contribution from weight, SAT, VAT, or total adipose tissue. CONCLUSION: In this middle-aged population, a negative relationship existed between MRI-measured BMAT and hip and lumbar spine BMD independent of demographics and body composition. These observations support the growing evidence linking BMAT with low bone density.