Abstract: In the process of designing the traditional extremum seeking control systems (ESCSs), the independent design method may lead the designed ESCSs to be difficult to get their optimal performance. Whereas, the current integrated design of ESCSs is an intricate method that needs to accomplish the integrated modeling for different controlled plants and the concrete extremum seeking algorithms (ESAs). To solve those problems, a universal integrated control method for block control ESCSs is proposed. Firstly, based on feedback linearization technology, the pseudo virtual control law is designed for block control ESCSs. Secondly, with the search variables of ESAs as the inputs, multilayer neural networks (MNNs) are used to approximately cancel the error between an approximate model and real model, and approximate the extrema of the states and the differential of the extrema. At the same time, an adaptive parameter and continuous robust terms are adopted to minify the influence of the MNNs construction error. Lastly, the virtual control laws and the final control law are derived using backstepping control method. This integrated control scheme guarantees that all of signals, which include the tracking errors of the states, the errors between the outputs and their extrema, the tracking errors of the search variables, and the estimation errors of MNNs' parameters, are the ultimate boundedness of the closed-loop system. Theoretical analysis and simulation results show the validity of this integrated control method.