In the dynamic action potential (AP) clamp, the membrane potential is forced to follow that of a real or modelled AP and membrane currents are measured (1). We apply the technique to different computational models describing cardiac cell electrophysiology to evaluate the relative roles of ionic current components in models with different AP shapes. This process is illustrated by identifying the ionic mechanisms for the different effects of rapid and slow delayed rectifier current (I_Kr and I_Ks) block on the formation of early afterdepolarisations (EADs) in human and guinea pig ventricular midmyocardial cell models. EADs can occur in the Luo-Rudy dynamic (LRd) guinea pig model (2) with 40% block of I_Kr or 70% block of I_Ks (3), but in the Ten Tusscher and Panfilov (TP) human model (4) 93% block of both I_Kr and I_Ks is required for EAD formation (5). This could simply be due to the different AP shapes, or may be a consequence of differing contributions of different currents to AP repolarisation and/or to different reactivation kinetics of the L-type Ca²⁺ current I_Ca,L that is responsible for EAD development. We therefore used the dynamic AP clamp technique to impose the same waveforms (either LRd guinea pig or TP human solitary APs) on each of the two models. During both guinea pig and human AP clamps, the repolarising currents I_Kr and I_Ks, the Na⁺-Ca²⁺ exchanger current I_NaCa, and the inward rectifier current I_K1, had the same relative contributions to the total repolarising current in both models. However, the inactivation gates of I_Ca,L recovered from inactivation more slowly in the TP human model, compared to the LRd guinea pig model. Regardless of waveform, the I_Ca,L voltage-dependent inactivation gate f had recovered to f ≈ 0.86 in the LRd model at time of 90% AP repolarisation, but to only f ≈ 0.46 in the TP model. The f gate recovers to f = 0.4 at approximately 84% and 95% of AP duration in the LRd and TP models respectively (again, regardless of waveform). Therefore, a relatively longer prolongation of the AP is required in the TP human model (313 ms or 75%, compared to 77 ms or 35% in LRd) for sufficient reactivation of I_Ca,L and subsequent EAD formation to occur. We conclude that the reduced propensity for EAD formation in the TP human model, compared to the LRd guinea pig model, is due primarily to the slower kinetics of I_Ca,L recovery from inactivation, and that a greater block of I_Kr and I_Ks is required only to sufficiently prolong the AP. We propose that the dynamic AP clamp can be utilised to compare other aspects of normal and pathological electrophysiology between different cardiac cell models.