An adaptive fuzzy controller is developed for AC servo system to guarantee both global stability and performance. AC motor dynamics under consideration may include large nonlinear uncertainties, such as nonlinearload variations. The Controller Output Method (COEM) isintroduced and applied to the design of adaptive fuzzycontrol system. The method employs a gradient descentalgorithm to minimize a cost function that is based on theerror of the controller output. The cost function can beminimized through adapting some or all of the parameters of fuzzy controller. The proposed adaptive fuzzy controlleris applied to a AC servo control problem and comparedwith a standard fuzzy controller. The simulation resultsshow that the proposed control is effective and yieldssuperior performance.