The amount of lubricating oil in a gear reducer greatly influences components’ temperature and efficiency of the reducer. A thermal network simulation model of a gear reducer for electric vehicles based on amesim software was built for thermal and efficiency analysis of the reducer, where heat production and transfer processes of each component in the reducer are considered. Moreover, secondary development submodels were developed to calculate the convection coefficients in real time based on different lubricating oil temperatures and working conditions. Next, the experiments were conducted to verify the developed model regarding transmission efficiency and lubricating oil equilibrium temperature in the reducer. The deviation between the numerical and experimental results is less than 5%, which indicates high accuracy of the developed thermal network model. Thereafter, efficiency and heat balance simulation tests under different working conditions were performed to determine an optimal lubricating oil supply, which is 14 mm gear-oil-immersion oil supply in this study. Finally, a comparison between the reducer with the optimal oil supply and that with normally used 22 mm gear-oil-immersion supply in the NEDC cycles in low temperature was done. The reducer with the optimal oil supply shows 1.032% energy saving. The method presented in this article provides a good guide to design and optimization of the gear reducer for electrical vehicles.