Aiming at the difficulty of modeling thermal state characteristics and identifying thermal key points in the thermal error compensation technology of machine tools,a more complete set of spindle thermal state modeling methods and rapid identification methods of thermal key points were proposed.The spindle thermal characteristics analysis model was established by considering the spindle system temperature and thermal deformation. The results show that the error between the predicted value of the model and the measured value of a precision horizontal machining center is within 20%,which shows the correctness of the proposed modeling method. The output of the model was used to identify the thermal key points of the main shaft,and six thermal key points were obtained according to the heat transfer function values of 12 measured points; according to the data of six key points,a main shaft thermal error prediction model was established by using the BP neural network. The predicted output values of the BP neural network were compared with the output values of the thermal characteristic model. The results show that the maximum error is -0.060 36 μm and the maximum relative error is -0.200 6%,which verifies the effectiveness of the proposed method for identifying thermal key points of heat.