This research is concerned with a compensation method of a mobile manipulator using a neural network for moving operation and with the tracking control performance. The mobile manipulator can spread its working area widely and improve its working efficiency. Several papers of mobile manipulators have been presented and assumed that the traveling terrain is flat. But, irregularities exist even on an artificial floor in factories and buildings, and it is difficult to measure exactly the shape of the terrain. When the end-effector of the mobile manipulator tracks desired trajectory during the mobile manipulator travels on an unknown irregular terrain, it is subjected to the effect of the disturbance torques derived from influences depending on the terrain. Furthermore, the disturbance torque derived from the change of payload disturbs the accurate trajectory control. Then, a compensation method for the two kinds of the disturbance torque must be considered. First, for the change of the payload, the adaptive controller is constructed for the mobile manipulator. Second, the control performance of the adaptive controller is examined. Third, for the disturbance depending on the terrain, the compensator using a neural network is proposed. Finally, the validity of the adaptive controller with a neuro compensator is clarified through simulation experiments.