<p>Accurately simulating the interannual variability of the East Asian winter monsoon (EAWM) using climate models remains challenging, with uncertainties in physical parameterization being one of the key constraints on simulation performance. This study quantitatively investigates the contributions and mechanisms of model physical parameters to the interannual variability of the EAWM based on a 40-year-long Perturbed Parameter Ensemble (PPE) in CAS-FGOALS-g3. We find that simulations of the interannual variability of the EAWM show diversity in the presence of parameter perturbations, and that the vast majority of the variability can be explained by a small group of parameters from convection and cloud microphysics. Parameter perturbation mainly alters the strength of tropical anomalous convection and the atmospheric circulation response to El Niño-Southern Oscillation (ENSO) that are the most critical factors affecting the simulation of the EAWM’s interannual variability. We carry out a further explanation with the highest sensitivity parameter, namely evaporative efficiency of the deep convection. An increase in evaporative efficiency results in higher low-level water vapor concentrations in the central and eastern equatorial Pacific, amplifying local convective anomalies and weakening the Walker circulation. Consequently, this leads to enhanced anticyclonic anomalies in the lower troposphere of the Western North Pacific, thereby intensifying the interannual variability of the EAWM, and vice versa. These insights enhance our understanding of how tropical teleconnections influence the EAWM’s interannual variability and can help reduce simulation uncertainties by calibrating and improving the most influential parameterizations, namely deep convection in this study.</p>