Background/Objectives: There is no consensus on the definition of sarcopenic obesity (SO), resulting in inconsistent associations of SO with mortality risk. We aim to evaluate association of dual energy x-ray absorptiometry (DXA) SO models with mortality risk in a US adult population (≥50 years). Subjects/Methods: The study population consisted of 3577 participants aged 50 years and older from the 1999–2004 National Health and Nutrition and Examination Survey with mortality follow-up data through December 31, 2011. Difference in survival time in people with and without SO defined by three body composition DXA models (Model 1: body composition phenotype model; Model 2: Truncal Fat Mass (TrFM)/Appendicular Skeletal Muscle Mass (ASM) ratio model; Model 3: Fat Mass (FM)/Fat Free Mass (FFM) ratio). The differences between the models were assessed by the acceleration failure time model, and expressed as time ratios (TR). Results: Participants age 50–70 years with SO had a significantly decreased survival time, according to the body composition phenotype model (TR: 0.92; 95% CI: 0.87–0.97), and TrFM/ASM ratio model (TR: 0.88; 95% CI: 0.81–0.95). The FM/FFM ratio model did not detect significant differences in survival time. Participants with SO aged 70 years and older did not have a significantly decreased survival time, according to all three models. Conclusions: A SO phenotype increases mortality risk in people of age 50–70 years, but not in people aged 70 years and older. The application of the body composition phenotype and the TrFM/ASM ratio models may represent useful diagnostic approaches to improve the prediction of disease and mortality risk.