An approach is developed using dynamic programming to obtain an optimal pricing policy for chartered flights. Since the dynamic programming model has a very large dimension in real world cases, Q Learning technique is used to develop an approximate solution method. Analysis is carried out assuming either linear-deterministic or probabilistic demand, with an exact solution found for deterministic demand and a Q Learning approach proposed for probabilistic demand that also considers reservation prices. The method is implemented using real sales data from Tehran to Mashhad flights and shown to perform well for deterministic demand with enough iterations and simulate high and low travel seasons for probabilistic demand when using reservation prices.