A systematic approach to identifying Hammerstein and Wiener models, including the model structure and parameters, for nonlinear processes is presented. A sequence of predesigned relay magnitude changes is applied in a relay feedback experiment to generate dynamic response for identification. By way of this proposed approach, the identifications of linear and nonlinear elements in the block-oriented models are fully separated and the needs for iterative optimization procedures can be relaxed. With the nonlinear models obtained, nonlinear control strategies are developed to achieve a better control performance. The effectiveness of the proposed identification method and control strategy is demonstrated with simulated examples.