Based on the regional climate model RegCM4.5 driven by National Centers for Environmental Prediction (NCEP) reanalysis, the influence of cumulus convection schemes (CCSs) on the winter North Pacific storm track (WNPST) is investigated. It is found that the climatology, interannual variation, spatial modes and characteristic indices of the WNPST are extremely sensitive to the choice of CCS. Among the selected CCSs, WNPST climatology and interannual variation in the Kuo scheme are better than in other CCSs, with a smaller root mean square error. The WNPST spatial modes and strength indices in the Kuo and Grell schemes are more consistent with NCEP reanalysis. The Kuo scheme has a stronger ability to simulate the WNPST latitude index and the interannual variation of winter characteristic indices. In addition, we attempt to reveal the possible reasons for the different performances of CCSs from the viewpoint of baroclinic energy conversion (BCEC). It is found that the energy conversion from the mean available potential energy to the eddy available potential energy (BCEC1) has no significant difference among the Kuo, Grell and Emanuel schemes, while energy conversion from the eddy available potential energy to eddy kinetic energy (BCEC2) in the Kuo scheme is obviously better than other CCSs, which means that the differences in BCEC2 among these CCSs may be one of the key reasons affecting the simulation results of the WNPST.