HERG (human ether-a-go-go related gene, Kv11.1, KCNH2) is a voltage-gated potassium channel with unique gating characteristics. HERG has fast voltage-dependent inactivation, relatively slow deactivation, and fast recovery from inactivation. This combination of gating kinetics makes study of HERG difficult without using mathematical models. Several HERG models have been developed, with fundamentally different organization and properties.We programmed five distinct HERG models and tested their behavior under voltage-clamp and guinea-pig ventricular myocyte action potential clamp. Four models used Markov formalisms, and one used Hodgkin-Huxley formalism. HERG behavior cannot be replicated using a Hodgkin-Huxley formalism. The Markov Models had 2 or 3 closed states, 1 open state, and 1 inactivated state. A voltage-independent activation step is required in order to replicate the experimentally observed voltage-independent rate limiting step of activation. A fundamental difference between models is the presence or absence of a transition directly from the closed state to the inactivated state. Two of the Markov models had transitions directly from the closed state to the inactivated state. Our analysis demonstrates that the only models which effectively reproduce HERG experimental data require that the closed-inactivated transitions are absent or are effectively zero compared to the closed to open transitions, rendering the closed-inactivation transition superfluous. Furthermore, the models make significantly different predictions about the behavior of the HERG during action potentials and premature stimuli.Our simulations indicate there is no direct transition between the pre-activated closed state and the inactivated state, although inactivation from a “flicker” open state is possible. The use of a model with a direct transition between closed and inactivated states with rates that are effectively zero is potentially misleading in understanding HERG gating.