This paper presents a model reference adaptive control approach for the synchronization of a discrete-time chaotic systems using output tracking control. Takagi-Sugeno (T-S) fuzzy model is employed to represent the discrete-time chaotic system. Then, the reference model system is chosen using the output of master system and design the control input so that the controlled slave system achieves asymptotic synchronization with the reference system given that two systems start from different initial conditions, different parameters and/or different type of model. Using a gradient algorithm, the ideal controller gains which can stabilize the error equation are estimated. Simulation examples of two cases are given to demonstrate the validity of our proposed adaptive method.