A desired compensation adaptive robust control (DCARC) approach is proposed for nonlinear systems having both parametric uncertainties and uncertain nonlinearities. DCARC of nonlinear systems transformable to a normal form is first solved. A DCARC backstepping design is then developed to overcome the design difficulties associated with unmatched model uncertainties. The proposed DCARC has the unique feature that the adaptive model compensation part depends on the reference trajectory and parameter estimates only. Such a structure has several implementation advantages. First, the regressor in the model compensation part can be calculated off-line and on-line computation time may be reduced. Second, the interaction between the parameter adaptation and the robust control law is minimized, which may facilitate the controller gain tuning process considerably. Third, the effect of measurement noise is minimized since the regressor does not depend on actual measurements. As a result, a fast adaptation rate may be chosen in implementation to speed up the transient response and to improve overall tracking performance. These claims have been verified in the comparative experimental studies for the control of robot manipulators.