Background: Dynamic action potential (AP) clamp (dAPC) is an electrophysiology technique that allows one to study in real time the effects of a biological current included in a computational AP model.dAPC improves on classical electrophysiology techniques as it enables observation of the effects of a drug on the current of interest, and on the AP at the same time.During an experiment, however, the seal resistance between the cell membrane and the pipette is finite and a leak current occurs.Its reduction is crucial to assess the drug effect correctly. Aim: Our work aims to quantify the impact of the leak current on a ventricular AP model and to evaluate the benefits of an online compensation. Methods: Experiments were performed using a prototype Nanion Dynamite8 system in dAPC mode running the ten Tusscher (TP06) human ventricular AP model.We used a passive model cell (Cm=19.8 pF,Rseal=500 MOhm) and downscaled its current by a factor x0.1 to keep the AP model stable. We compensate online the leak current adopting a linear model.APD90 and RMP were measured at several degrees of compensation (0,25,50,75,100%) and at different pacing frequencies (0.5,1,2Hz), and compared with the AP model in the open loop (i.e. with no connection between the model cell and the AP model). Results: APD90 showed minimal variations (<1%) at 0.5Hz, whereas the impact of leak was higher at 1 and 2Hz (-3.6% and -4.1% with no compensation).The compensation was found to be more critical for RMP.The leak current was responsible for depolarization of RMP up to 7.2% at 2Hz. Conclusion: With this test, we showed that online compensation of the leak current is beneficial for proper assessment of APD90 and RMP.This aspect is especially critical for higher pacing frequencies, as these showed larger variations from the uncoupled real time computational AP model.