This study was aimed at determining the correlation between regional body composition and total body fat percentage (TBF%), as measured by dual-energy X-ray absorptiometry (DXA), among Chinese women and men. Men (n=203) and women (n=630), aged 20-87yr, were assessed for height, weight, and body mass index (BMI) and for total body, lumbar spine (L2-L4), and left and right proximal femur fat percentage using DXA. Based on their BMI, male and female subjects were subdivided into 4 groups. For all groups, the TBF% derived from the total body scan was strongly correlated with spine %fat (r range: 0.60-0.96) and the left and right femur %fat (r range: 0.72-0.80 and 0.69-0.83, respectively), as obtained from regional scans. In stepwise regression, TBF% was best assessed by spine %fat, left femur %fat, age, and BMI for both genders (male: R(2)=0.88; female: R(2)=0.84). Equations with spine %fat, age, and BMI also provided good estimates of TBF% (male: R(2)=0.80; female: R(2)=0.75), whereas android and gynoid %fat were predicted with spine %fat and left femur %fat, respectively, and the predicted fat percentage ratio of android to gynoid (A/G) could be obtained from spine %fat and left femur %fat. The cross-validation results showed that the accuracy of all prediction equations was high in the male group, whereas prediction equations with high-to-moderate accuracy were found in the female group. Therefore, for DXA systems that do not offer total body scan, clinicians can assess TBF% using data obtained from regional scans.