Objectives: The purpose of this study was to determine the relationship between body composition, N-terminal B-type natriuretic peptide (NT-proBNP) levels, and heart failure (HF) phenotypes and outcomes. Background: Abnormalities in body composition can influence metabolic dysfunction and HF severity; however, data assessing fat distribution and skeletal muscle (SM) size in HF with reduced (HFrEF) and preserved EF (HFpEF) are limited. Further, whether NPs relate more closely to axial muscle mass than measures of adiposity is not well studied. Methods: We studied 572 adults without HF (n = 367), with HFrEF (n = 113), or with HFpEF (n = 92). Cardiac magnetic resonance was used to assess subcutaneous and visceral abdominal fat, paracardial fat, and axial SM size. We measured NT-proBNP in 334 participants. We used Cox regression to analyze the relationship between body composition and mortality. Results: Compared with controls, pericardial and subcutaneous fat thickness were significantly increased in HFpEF, whereas patients with HFrEF had reduced axial SM size after adjusting for age, sex, race, and body height (p < 0.05 for comparisons). Lower axial SM size, but not fat, was significantly predictive of death in unadjusted (standardized hazard ratio: 0.63; p < 0.0001) and multivariable-adjusted analyses (standardized hazard ratio = 0.72; p = 0.0007). NT-proBNP levels more closely related to lower axial SM rather than fat distribution or body mass index (BMI) in network analysis, and when simultaneously assessed, only SM (p = 0.0002) but not BMI (p = 0.18) was associated with NT-proBNP. However, both NT-proBNP and axial SM mass were independently predictive of death (p < 0.05). Conclusions: HFpEF and HFrEF have distinct abnormalities in body composition. Reduced axial SM, but not fat, independently predicts mortality. Greater axial SM more closely associates with lower NT-proBNP rather than adiposity. Lower NT-proBNP levels in HFpEF compared with HFrEF relate more closely to muscle mass rather than obesity.