A novel intelligent control scheme is presented for accurate position tracking of electrohydraulic servo actuators. The proposed control law is designed by means of a non-linear control approach and includes an adaptive neural network to provide the basic intelligent features. Online learning, instead of off-line supervised training, is proposed to update the weight vector of the neural network. Moreover, the adoption of a composite error signal as the only input to the neural network allows a significant reduction in the computational complexity of the algorithm. Rigorous proofs for the boundedness and convergence properties of the closed-loop signals are provided. Experimental results obtained with an electrohydraulic system demonstrate the efficacy of the proposed controller, even considering the highly non-linear and uncertain plant dynamics.