In this paper, we analyze forecast based inventory control policies for a non-stationary demand. We assume that forecasts and the associated uncertainties are given at the beginning of the horizon of forecasts. Two forecast based reorder point policies are proposed : the (rk;Q) and the (rk;Qk) policies. These dynamic policies represent an extension of the classical discrete time (r;Q) policy. The parameters of these policies are determined by using a sequential approach which satisfies a cycle service level. A numerical comparative study of these policies is developed enabling us to show the benefit of using them when forecasts are reliable.