Abstract: A third generation Benchmark problem on structural control is investigated and a controller based on fuzzy neural network is designed. Numerical simulation is carried out for analyzing the nonlinear seismic responses of the controlled 20-story Benchmark building with MR (Magneto-Rheological) damper. Firstly, a neural network is employed to identify the dynamic characteristics of the full-scale MR damper. In designing a fuzzy neural network controller, a new idea based on zoning control is proposed. The simulation results are compared with those of LQG sample-control systems. The analytical results show that the developed semi-active strategy has good performance, and can reduce the nonlinear seismic response of high-rise building and the damages in the building structures caused by strong earthquakes. Although the control efficiency of the intelligent controller is a little less than that of LQG-sample active controller on some values of evaluation criteria, the intrinsic robustness of controller can prove that the intelligent controller has a good potential in the control of nonlinear vibrations of structures.