The authors present an optimization strategy applied to decelerating cascades, with the combined use of a Navier-Stokes flow solver, correlation functions and a gradient based optimization method. In the initial stages of the searching the configurations proposed by the optimizer are mainly investigated with a semi-empirical analysis tool so to provide the flow solver with an improved initial guess. In the later stages the optimizer directly controls the flow solver. The objective of the optimization is two fold: determining the geometrical characteristics of the cascade yielding the best aerodynamic performances, and defining an appropriate cost function accounting for optimality conditions in a more general sense. The method is applied to blade profiles of the C4 type, whose geometrical characteristics are determined as a function of a few parameters (typically the camber angle, the maximum thickness to chord ratio, and the chord). By doing so the number of design parameters is substantially reduced, and the validity of the present methodology is correctly demonstrated, without loss of generality, with a limited computational effort. The examples deal with the design of decelerating cascades realizing considerable flow turning both in subsonic and transonic regimes and demonstrate the potential of the method.