Abstract: In this study, a composite adaptive sliding mode control using Multiple Models (MM-CASMC) is proposed for precision position control of an induction motor servo system with parametric uncertainties and external disturbance. The MM-CASMC is designed based on a classical sliding mode control frame. Robustness against parametric uncertainties and high-frequency extern disturbance are both obtained via online parameters estimation and switching control, respectively. A composite adaption law which combines direct and BGF-LS type indirect adaptive methods is developed to achieve both Globally Uniformly Ultimately Boundness (GUUB) and approximately exponential convergence in large range under persistent excitation, the later implies clearer transient behaviour which is of great importance but not provided by standard direct adaptive method. Moreover, a multiple model adaptive control design is further incorporated to achieve improvement in transient response by utilizing model switching and parameters estimates resetting and an noval method by means of dual-channel filtering is proposed for regessor filtering and model switching. For the proposed strategy, the GUUB stability and improvments in transient behaviour and adaptability to sudden changes in the parameter values are all proved in Lyapunov sense. Simulation results verify that an induction motor servo system with the adoption of MM-CASMC can achieve favorable tracking performance and transient response in the presence of parameter variations and external load disturbance.