In this paper, we propose an observer-based fuzzy adaptive control scheme for a class of multiple- input-multiple-output (MIMO) nonlinear non-affine systems with uncertainty and unknown state derivative. The proposed approach combines the recurrent-type fuzzy logic system (RFLS) and nonlinear observer to deal with the tracking control problem. It adopts the RFLS to approximate nonlinear unknown functions and the nonlinear observer is established to estimate the unmeasured derivative states. Based on the Lyapunov stability approach, the update laws of RFLS are obtained and the stability of the closed-loop system is guaranteed. Finally, our approach is applied in tracking control of a two-input-two-output system and two-link manipulator robot system. Simulation and experimental results of manipulator robot are presented to show the effectiveness and performance of the proposed control scheme.