This paper proposes a simple iterative method - time iteration - to solve linear rational expectation models. I prove that this method converges to the solution with the smallest eigenvalues in absolute value, and provide the conditions under which this solution is unique. In particular, if conditions similar to those of Blanchard and Kahn (1980) are met, the procedure converges to the unique stable solution. Apart from its transparency and simplicity of implementation, the method provides a straightforward approach to solving models with less standard features, such as regime switching models. For large-scale problems the method is 10-20 times faster than existing solution methods.