In this paper we present a method for using rational expectations in a linear-quadratic optimizationframework. Following the approach put forward by Sims, we solve the model through a QZdecomposition, which is generally easier to implement than the more widely used method of Blanchardand Kahn.