A new dc motor control technique for the Coulomb friction compensation is proposed. The technique uses an adaptive velocity control scheme for a dc servo motor with on-line estimated parameters, including a Coulomb friction parameter, which is a combination of the Coulomb friction torque, motor time constant, moment of inertia of the motor, and sampling time of the discrete-time motor model. The estimation model used in the adaptive control process is validated off-line by a pseudo-linear regression algorithm for system parameters in a linear ARMAX model, and by adaptive Kalman filters for the Coulomb friction parameter described as pseudo-random binary sequences. The adaptive controller consists of a friction compensator and a PID controller, whose gains are adjusted adaptively in terms of estimated parameters. The proposed adaptive control law is implemented and tested on a microprocessor-based dc servo motor, and is applicable to many dc-motor-driven precision servo mechanisms. Experimental results are shown to be superior to those of conventional PID controls in terms of parameter fluctuation.