This paper is concerned with the identification of nonlinear systems by utilizing of hybrid soft computing approaches. Based on the flexibly computational structure of the tree, a unified framework is constructed in which various soft computing models can be developed, evolved and evaluated. In this framework, the architecture of the hybrid soft computing models is created and evolved by using the modified probabilistic incremental program evolution (MPIPE) algorithm, and the parameters used in hybrid soft computing models can be optimized using a class of optimization techniques. Simulation results for the identification of nonlinear systems show the feasibility and effectiveness of the proposed method.