Abstract An iterative technique for the derivation of a nearly optimal feedback control law for continuous dynamic systems with separable cost functions is developed based on the fundamental concepts of multilevel control. The developed technique comprises two major and subsequent stages. In the first stage, a two-level optimization structure is devised using the coordination by control method. An identification processthen follows in order to determine the constant gains of the dynamic controller utilizing the discrete quasilinearization algorithm. A three-level control structure is thus provided for the overall technique which can adaptively compute the unknown gains of the dynamic controller.