Robust recurrent-neural- network(RRNN) sliding-mode control may track biaxial motion mechanism of reference contour. The biaxial motion mechanism is a computer-controlled X-Y table numerical machine that is driven by two vector contro1led permanent-magnet synchronous motors. A RRNN sliding-mode control system is based on the excellent dynamic characteristics derived from controlling X-axis and Y-axis respectively. The motion tracking performance is significantly improved through employing robustness control techniques of parameter variations, external disturbances, cross coupled interference and frictional torque. Experimental results on circular reference contours show that the dynamic behaviour of the anti-disturbance of the proposed control system reaches without considering existing uncertainties.