Abstract: In thermal balance test of the spacecraft, using temperature extrapolation technology obtain the ultimate thermal balance temperature of the spacecraft can be predicted accurately and the test duration can be shortened, and the development cost of the spacecraft can be reduced accordingly. The nonlinear least-squares estimation and the extended Kalman filtering (EKF) are two of the most pervasive extrapolation and prediction methods, some important engineering problems about how to apply these two methods in spacecraft thermal balance test are discussed.