Left ventricular (LV) stroke work (SW) is calculated from the pressure-volume (PV) loop. PV loops do not contain information on longitudinal and radial pumping, leaving their contributions to SW unknown. A conceptual framework is proposed to derive the longitudinal and radial contributions to SW, using ventricular force-length loops reflecting longitudinal and radial pumping. The aim of this study was to develop and validate this framework experimentally and to explore these contributions in healthy controls and heart failure patients. Thirteen swine underwent cardiovascular magnetic resonance (CMR) and LV pressure catheterization at baseline (n = 7) or 1 wk after myocardial infarction (n = 6). CMR and noninvasive PV loop quantification were performed on 26 human controls and 14 patients. Longitudinal and radial forces were calculated as LV pressure multiplied by the myocardial surface areas in the respective directions. Length components were defined as the atrioventricular plane and epicardial displacements, respectively. Contributions to SW were calculated as the area within the respective force-length loop. Summation of longitudinal and radial SW had excellent agreement with PV loop-derived SW (ICC = 0.95, R = 0.96, bias + SD = = 4.5 + 5.4%) in swine. Longitudinal and radial contributions to SW were ~50/50% in swine and human controls, and 44/56% in patients. Longitudinal pumping required less work than radial to deliver stroke volume in swine (6.8 + 0.8 vs. 8.7 + 1.2 mJ/mL, P = 0.0002) and in humans (11 + 2.1 vs. 17 + 4.7 mJ/mL, P < 0.0001). In conclusion, longitudinal and radial pumping contribute ~50/50% to SW in swine and human controls and 44/56% in heart failure patients. Longitudinal pumping is more energy efficient than radial pumping in delivering stroke volume.