This paper develops a method for choosing advertising media plans for the next T-periods in order to maximize net discounted profits. Using the discrete maximum principle of Optimal Control Theory, it is shown that under certain conditions a T-period media model can be decomposed into a sequence of T one-period models together with the determination of the corresponding carry-over effects using an iterative procedure. The single period models with their objective functions modified to include the carry-over effects can be solved by conventional methods. Although the conditions for decomposition are not always satisfied, the heuristic solution procedure defined by the decomposition approach has been found to give close to optimal solutions in randomly generated test problems. The solutions obtained in these tests are usually better than the corresponding solutions obtained by the Little and Lodish type of heuristic applied to the present model. Finally, the T-period model is specialized to derive a steady-state formulation of an infinite horizon media planning problem.