In this paper, an intelligent controller is developed for hybrid force and position control of robot manipulators in the presence of external disturbances and the model uncertainties. The proposed controller consists of a model based controller and neural network based model free controller with an adaptive bound part. A non linear function of model dynamics is identified by employing a radial basis function neural network. The role of adaptive bound part is to estimate the bounds on model disturbances, friction term and neural network reconstruction error. The Lyapunov function candidate is used to prove the stability of the proposed controller and to show that the errors are asymptotically convergent. Finally numerical simulation results are presented for two link robot manipulator to show excellent performance of the proposed controller in comparison to other control schemes such as model based computed torque control and neural network based model free controller.