It is well known that extracellular calcium concentration ([Ca2+]o) affects cardiac action potential duration (APD): their inverse relationship has been experimentally observed in vivo and in vitro. Both shortening and prolongation of action potential (AP) are associated with an increased risk of arrhythmias and extracellular calcium variations may occur in many clinical contexts. Computational modeling could be a useful tool to explore this phenomenon: however, most of the commonly used ventricular cell models are not able to reproduce [Ca2+]o effects on AP duration properly, i.e. they respond to [Ca2+]o increase with AP prolongation. The aim of this study has been to modify the most recent human ventricular AP model, in order to improve its response to [Ca2+]o variations.The O'Hara-Rudy AP model (ORd) has been used as basis. Its L-type calcium current formulation has been replaced by a novel Markov model. Calcium-dependent inactivation (CDI) has been strengthened, with respect to the Voltage-dependent one (VDI). All inactivation rates are actually voltage-dependent: Ca2+ modulates only the fraction of channels working in CDI mode, which is 10 times faster than the VDI. After few minor changes, all involving Ca2+-handling, the modified model has been validated against a wide set of experimental data. Simulations were run with the original and modified models considering [Ca2+]o ranging from 1 to 3 mM.In control ([Ca2+]o =1.8mM), the modified and original ORd model APs and ionic currents were very similar. However, when changing [Ca2+]o from 0.9 to 2.7 mM, APDs varied in an opposite way: +19% vs −12%, original and modified models respectively. Therefore, the modified model may be used to explore in silico electrolyte changes effects on AP. Moreover, these results suggest that CDI mechanism is usually underestimated in ventricular human AP models.