Since the characteristics of time-delay, nonlinear and uncertain model structure, and In order to control temperature in greenhouse better, a fuzzy neural network PID control is presented in this paper. The intelligent controller can adjust the parameters according to the variation of system characteristics even if the system model is unknown, and meet the requests of real-time control. The approach shows the characteristics of outstanding adaptability and self-study ability, satisfactory dynamic performance, and certain application values by simulation results.