According to the balance of plant canopy energy, a mathematical model of simulating greenhouse plant canopy temperature that has relations with greenhouse inside and outside climate and structure was developed. The experiment was carried out to validate the correctness of the model in the Venlo-type greenhouse in North China during winter. The simulated canopy temperature agreed well with the measured data. The determination coefficient and the root mean square error between the measured and simulated canopy temperature were 0.7975 and 1.3℃, respectively. Then a BP neural network model of canopy temperature was built, the determination coefficient and the root mean square error were 0.7835 and 0.6℃. Based on the neural network model, the influencing factors sensitivity of canopy temperature was analyzed. The inside temperature is the most important factor, and the next ones are in order of the transpiration rate, solar radiation and inside relative humidity.