In order to eliminate nonlinear effects of friction on 2-DOF tracking system and to improve its position-tracking accuracy, a novel adaptive friction compensation algorithm is proposed. Based on nonlinear characteristics of friction, a LuGre dynamic friction model is adopted to characterize friction in the 2-DOF tracking system. In this method, unknown parameters for the LuGre model are estimated in the manner of off-line; friction state and the inconstant parameters are estimated by using nonlinear observers and self-adaptive law. A Lyapunov-like argument proves the asymptotic stability of the whole closed-loop system. Experimental results indicate that the proposed method could effectively alleviate the influences of friction when the 2-DOF tracking system works with a lower speed, and could improve tracking accuracy for the 2-DOF tracking system accordingly.