© 2016 RSNA. Purpose: To determine if computed tomographic (CT) metrics of bone mineral density and muscle mass can improve the prediction of noncancer death in men with localized prostate cancer. Materials and Methods: Institutional review board approval was obtained, with waiver of informed consent. All patients who underwent radiation therapy for localized prostate cancer between 2001 and 2012 with height, weight, and past medical history documented and who underwent CT that included the L4-5 vertebral interspace were included. On a single axial CT section obtained at the mid-L5 level, the mean CT attenuation of the trabecular bone of the L5 vertebral body (L5 HU ) was measured. The height-normalized psoas cross-sectional area (Psoas L4-5 ) was measured on a single CT section obtained at the L4-5 vertebral interface. Multivariable Cox proportional hazards models were used to assess effects on noncancer death. By using parameter estimates from an adjusted model, a prognostic index for prediction of noncancer death was generated and compared with age-adjusted Charlson Comorbidity Index (CCI) by using the Harrell c statistic. Results: Six hundred ffty-three men met the inclusion criteria. Prostate cancer risk grouping, androgen deprivation, race, age-adjusted CCI, L5 HU , and PsoasL4-5 were included in a multivariable model. Age-adjusted CCI (hazard ratio [HR] = 1.36, P <.001), L5 HU (HR = 2.88 for L5 HU , 105 HU, HR = 1.42 for 105 HU L5 HU 150 HU, P <.001), Psoas L4-5 (HR = 1.95 for PsoasL4-5, 7.5 cm 2 /m 2 , P =.003), and race (HR = 1.68 for African American race, HR = 1.77 for other nonwhite race, P =.019) were independent predictors of noncancer death. The prognostic index yielded a c value of 0.747 for the prediction of noncancer death versus 0.718 for age-adjusted CCI alone. Conclusion: L5 HU and PsoasL4-5, which are surrogates for bone mineral density and muscle mass, respectively, were independent predictors of noncancer death. The prognostic index that incorporated these measures with the CCI was associated with improved accuracy for prediction of noncancer death.