Reversible circuits, i.e. circuits which map each possible input vector to a unique output vector, build the basis for emerging applications e.g. in the domain of low-power design or quantum computation. As a result, researchers developed various approaches for synthesis of this kind of logic. In this paper, we consider the ESOP-based synthesis method. Here, functions given as Exclusive Sum of Products (ESOPs) are realized. In contrast to conventional circuit optimization, the quality of the resulting circuits depends thereby not only on the number of product terms, but on further criteria as well. In this paper, we present an approach based on an evolutionary algorithm which optimizes the function description with respect to these criteria. Instead of ESOPs, Pseudo Kronecker Expression (PSDKRO) are thereby utilized enabling minimization within reasonable time bounds. Experimental results confirm that the proposed approach enables the realization of circuits with significantly less cost.