Friction is an inevitable issue in most of mechanical servo systems. Friction has a counter-effect on the dynamic performance of servo mechanisms and needs to be considered in the design process. Particularly, in high-performance motion systems, friction can severely deteriorate performance and can cause tracking errors, longer settling time and limit cycles. In this paper, a new method is presented based on the adaptive fuzzy sliding mode control strategy for position control of mechanical systems. The first order dynamic LuGre model is used for the design of friction observer. Unlike previous papers, the control input and friction are applied to the system with non-equal gains. An adaptive law is employed for the estimation of the ratio between the gains of input and friction terms. Various uncertainties on parameters of friction also are considered and an appropriate control strategy is designed to tackle these uncertainties.