Both invasive left-ventricular pressure measurements and non-invasive colour M-mode echographic measurements have shown the existence of intraventricular pressure gradients (IVPGs) during early filling. The mechanisms responsible for these IVPG cannot be completely explained by the experiments. Therefore a one-dimensional numerical model is developed and validated. The model describes filling (both velocities and pressures) along a left ventricular (LV) base-apex axis. Blood-wall interaction in the left ventricle with moving boundaries is taken into account. The computational results for a canine heart indicate that the observed IVPGs during filling are the consequence of a complex interaction between, on the one hand, pressure waves travelling in the LV and, on the other hand, LV geometry, relaxation and compliance. The computational results indicate the pressure dependency of wavespeed (0.77-1.90 m-1 s) for different mean intraventricular pressures (0.88-5.00 mmHg) and IVPGs up to 2 mmHg, independent of the ratio of end systolic volume and equilibrium volume. Increasing relaxation rate not only decreases minimum basal pressure (2.8 instead of 3.6 mmHg) but also has a strong influence on the time delay between the minimum basal and apical pressures (14 ms instead of 49 ms). The results sustain the hypothesis that pressure-wave propagation determines IVPGs and that IVPGs are no proof of elastic recoil.