In this paper, an improved analytical model with thermal effect is described for an outer rotor surface permanent magnet machine used for in-wheel motor electric powertrain. As the in-wheel motor is operating at high torque density, it is important to estimate the winding and magnet temperatures accurately and update the winding resistance and magnet remanence at every operating point in the efficiency calculation. An electromagnetic model based on conformal mapping is used to compute the field solution. The air-gap geometry is mapped to a simpler slotless shape, where the field solution can be obtained by solving Laplace equation. The canonical domain solution is mapped back to the original domain and verified with finite element (FEM) results. The closed form solutions for core loss and magnet loss are derived from the air-gap field solutions. The copper loss is calculated by considering the proximity and skin effects. The temperature influence on each loss term is studied in detail at different operating conditions. A thermal model is built using a lumped parameter thermal network and an improved discretisation approach. The model has been validated experimentally by comparing the measured end-winding and coolant temperatures with the model estimations while exciting the armature windings with DC current and varying the cooling conditions. The optimum number of employed nodes ensures a good trade-off between accuracy and computational effort. The electromagnetic, thermal, and mechanical models are combined to evaluate the powertrain efficiency. The energy consumption calculation over NEDC driving cycle is performed and the benefit of including thermal effect on the accuracy of the model has been quantified. Thus, the paper proposes a tool to obtain the motor efficiency and temperatures accurately with less computational effort. All the models are validated using Motor-cad software.