In order to realize the temperature prediction for nano satellite on orbit, a basic thermal environment model of nano satellite was established and analyzed. A BP neural network was used to predict the temperature of nano satellite based on it. After the analysis of the thermal environment model, the thermal flow of satellite-s crust, radiater and equipments in cabin was got and used as the inputs of the BP neural network, their temperatures were used as the outputs and data sample of BP neural network was got and used to train BP neural network. The BP neural network which was trained is the temperature prediction model of nano satellite and used to predict the temperature of the satellite on orbit after 10��s. Results fit the actual temperature of nano satellite on orbit well, and show that the temperature prediction model of neural network can predict the temperature of nano satellite accurately and rapidly.