Recently there has been an increasing interest in primal-dual methods for model predictive control (MPC), which require minimizing the (augmented) Lagrangian at each iteration. We propose a novel first order primal-dual method, termed proportional-integral projected gradient method, for MPC where the underlying finite horizon optimal control problem has both state and input constraints. Instead of minimizing the (augmented) Lagrangian, each iteration of our method only computes a single projection onto the state and input constraint set. We prove that our method ensures convergence to optimal solutions at Op1{kq and Op1{k 2 q rate if the objective function is convex and, respectively, strongly convex. We demonstrate our method via a trajectory-planning example with convexified keep-out-zone constraints.