This paper presents an Off-line Gain Adjustment (OGA) approach to reduce contour error in 5-axis high speed machining. The proposed contour error formulation is based on the estimation of tool contact points and the OGA is inspired from the idea of Model Predictive Control (MPC). The control gains used in the position loop of servo drives are optimally adjusted off-line to reduce the contour error for the considered trajectory. The obtained gain profiles are computed preserving axis kinematic limitations, stability criterion of servo drives and the motor current constraints. The OGA is developed thanks to a validated machine simulator. The simulation results prove that the OGA reduces significantly the contour error in 5-axis high speed machining.