In high power density applications, oil-cooled electrical machines have become a research hotspot due to their outstanding heat dissipation performance. Efficient thermal analysis methods are crucial for optimizing the heat dissipation performance of oil-cooled motor. In this study, a three-dimensional lumped parameter thermal network (LPTN) model is developed to accurately characterize the unique heat transfer paths of oil-cooled motors. To precisely calculate the losses in various parts of the motor, an electromagnetic finite element model is established, yielding the copper loss, stator core loss, rotor core loss, and permanent magnet eddy current loss. These loss values are incorporated into the LPTN to enable temperature prediction of the electrical machines. To validate the temperature estimation performance of the proposed LPTN, a fluid-thermal coupled simulation model is constructed to calculate the electrical machine’s steady-state temperatures under four operating points, which are then compared with the LPTN results. The simulation results indicate that the proposed LPTN model exhibits a maximum temperature estimation error of −9.8 ℃ under peak current conditions, corresponding to a relative error of 12.4%. Nevertheless, the model demonstrates remarkable computational efficiency advantages.