A new method of fast identifying temperature rise characteristics for machine tool spindle was developed. The nonlinear prediction technology based on support vector machine regression can predict the spindle temperature rise curve for a long time by only using a short period of measurement time. The steady-state temperature and thermal equilibrium time can be easily obtained from temperature prediction. The thermal balance test in the vertical machining center was done, and the method was applied under different working conditions. When spindle was running at 5000r/min with room temperature of 18℃, the temperature rise of the selected point was identified in 48min while the time for obtaining the temperature rise curve from start-up of machine tool to temperature steady-state machine tool in practice can reach 400min. The root mean square error (RMSE) between estimated and measured temperature was 0.1848℃, and the error between estimated and measured steady-state temperature was 0.0392℃. The method can greatly shorten thermal balance test duration. The experimental research illustrated that the method of identifying spindle temperature rise characteristics was fit for thermal balance test. It proved that the novel method was effective and feasible. 