Age-related decline in cardiac function can be prevented or postponed by lifelong endurance training. However, effects of normal ageing as well as of lifelong endurance exercise on longitudinal and radial contribution to stroke volume are unknown. The aim of this study was to determine resting longitudinal and radial pumping in elderly athletes, sedentary elderly and young sedentary subjects. Furthermore, we aimed to investigate determinants of maximal cardiac output in elderly. Eight elderly athletes (63 ± 4 years), seven elderly sedentary (66 ± 4 years) and ten young sedentary subjects (29 ± 4 years) underwent cardiac magnetic resonance imaging. All subjects underwent maximal exercise testing and for elderly subjects maximal cardiac output during cycling was determined using a dye dilution technique. Longitudinal and radial contribution to stroke volume did not differ between groups (longitudinal left ventricle (LV) 52–65%, P = 0.12, right ventricle (RV) 77–87%, P = 0.16, radial 7.9–8.6%, P = 1.0). Left ventricular atrioventricular plane displacement (LVAVPD) was higher in elderly athletes and young sedentary compared with elderly sedentary subjects (14 ± 3, 15 ± 2 and 11 ± 1 mm, respectively, P < 0.05). There was no difference between groups for RVAVPD (P = 0.2). LVAVPD was an independent predictor of maximal cardiac output (R2 = 0.61, P < 0.01, β = 0.78). Longitudinal and radial contributions to stroke volume did not differ between groups. However, how longitudinal pumping was achieved differed; elderly athletes and young sedentary subjects showed similar AVPD whereas this was significantly lower in elderly sedentary subjects. Elderly sedentary subjects achieved longitudinal pumping through increased short-axis area of the ventricle. Large AVPD was a determinant of maximal cardiac output and exercise capacity.