A new optimal guidance law which can satisfy terminal position, velocity and attitude constraints is proposed for a vertical landing rocket based on time-to-go weighted cost function. The theoretical result can be resolved to proportional form for the convenience of engineering application with several assumption. Then a simulated predictor and corrector is designed to compensate guidance error influenced by modeling error and internal and external uncertainty and perturbations. Characteristics of theoretical solution is analyzed. Nonlinear simulation indicate that the proposed guidance law is able to guide the rocket to land on target position with multiple terminal constraints and the acceleration command can converge to zero.