The test-analysis of effects on machine tool of motorized spindle temperature risings, thermal deformations and environment temperature in summer and in winter especially was presented, in order to reduce effects on machining precision by thermal deformation. Also the thermal error modeling based on using back propagation (BP) neural networking Volterra nonlinear system was realized. The analysis results indicate the environmental temperature effected by thermal diffusivities from headstock is raised rapidly in summer; thermal diffusivities of the machine tool is fast and the temperature is rising visibly on headstock in winter, and the environment temperature is almost in constant. The thermal deformations are similar in rules, but have some little difference in values between in summer and in winter for a same machine. It is validated by experiments that the model has merit of high predicative precision, which provides reference for compensation of real time thermal error in CNC machine tool.