In this paper, an intelligent controller to control a smart base-isolated masonry structure is proposed. The smart base isolation consists of rubber bearings and a magnetorheological damper. This semi-active scheme takes advantage of the low power requirement of magnetorheological devices to achieve structural control. The controller is capable of mapping the behaviour of the controlled structure sequentially, using a self-organizing neural network which has time-varying learning rates. This neurocontroller is augmented by a sliding mode controller to account for the uncertainty on the semi-active damping forces. Wavelets are used as functions in the single hidden layer for their capability of localizing functions in the space and frequency domains. Results show that the intelligent controller is capable of rapidly mapping the inverse stateforce relationship, without any prior knowledge of the dynamic properties of the controlled structure. The displacement control performance is similar to the performance from an active hybrid strategy, and the inter-story displacement is efficiently minimized.