Xin Su Ning (XSN), a China patented and certified multi-herbal medicine, has been available in China since 2005 for treating cardiac ventricular arrhythmia including arrhythmia induced by ischemic heart diseases and viral myocarditis, without adverse reactions being reported. It is vitally important to discover pharmacologically how XSN as a multicomponent medicine exerts its clinical efficacy, and whether the therapeutic effect of XSN can be verified by standard clinical trial studies. In this paper we report our discoveries in a cellular electrophysiological study and in a three-armed, randomized, double-blind, placebo-controlled, parallel-group, multicenter trial. Conventional electrophysiological techniques were used to study the cellular antiarrhythmic mechanism of XSN. Data was then modeled with computational simulation of human action potential (AP) of the cardiac ventricular myocytes. The clinical trial was conducted with a total of 861 eligible participants randomly assigned in a ratio of 2:2:1 to receive XSN, mexiletine, or the placebo for 4 weeks. The primary and secondary endpoint was the change of premature ventricular contraction (PVC) counts and PVC-related symptoms, respectively. This trial was registered in the Chinese Clinical Trial Register Center (ChiCTR-TRC-14004180). We found that XSN prolonged AP duration of the ventricular myocytes in a dose-dependent, reversible manner and blocked potassium channels. Patients in XSN group exhibited significant total effective responses in the reduction of PVCs compared to those in the placebo group (65.85% vs. 27.27%, P < 0.0001). No severe adverse effects attributable to XSN were observed. In conclusion, XSN is an effective multicomponent antiarrhythmic medicine to treat PVC without adverse effect in patients, which is convincingly supported by its class I & III pharmacological antiarrhythmic mechanism of blocking hERG potassium channels and hNaV1.5 sodium channel reported in our earlier publication and prolongs AP duration both in ventricular myocytes and with computational simulation of human AP presented in this report.