Oxygen and carbon are the first and second abundant elements, respectively, in the human body by mass. Although many physiological and pathological processes are accompanied with alteration of total body oxygen (TBO) and carbon (TBC) masses, in vivo measurements of the two elements are limited. Up to now, almost all available information of TBC and TBO is based on in vivo neutron activation (IVNA) analysis which is very expensive and involves moderate radiation exposure. The aim of the present study was to develop and evaluate an alternative strategy for TBC and TBO estimation. Mechanistic models were derived for predicting TBC and TBO masses from dual-energy x-ray absorptiometry (DXA) and total body water (TBW). Twenty-eight adult subjects were studied. IVNA-measured TBC and TBO masses were used as the criterion. TBC masses predicted by DXA-alone and by DXA-TBW models were 20.8 ± 7.1 kg and 20.6 ± 6.8 kg, respectively, close to the IVNA-measured value (19.5 ± 6.3 kg). There were strong correlations (both with r > 0.95, P < 0.001) between the predicted and measured TBC masses. TBO masses predicted by DXA-alone and by DXA-TBW models were 46.0 ± 9.8 kg and 46.5 ± 9.9 kg, respectively, close to the IVNA-measured value (48.0 ± 10.4 kg). Correlations (both with r > 0.97, P < 0.001) were strong between the predicted and measured TBO masses. Bland-Altman analysis validated the applicability of DXA-based models to predict TBC and TBO masses. As both DXA and TBW dilutions are widely available, low-risk, low-cost techniques, the present study provides a safe and practical method for estimating elemental composition in vivo.