A novel adaptive control for structures subjected to seismic excitation is presented. The aim of the control is to provide a high stabilizing performance involving a limited computational burden while allowing for frequent update of the control decision to cope with the changes in the excitation characteristics. Consequently, the control is based on a computationally efficient solution to the infinite-horizon linear optimal control problem, which employs the autoregressive model for excitation signals and the alpha-shift method for a performance index. Based on numerical simulations involving an actively controlled 20-story building subjected to different earthquake scenarios, we demonstrate that the adaptive control outperforms the standard LQG and H ∞ regulators. Our analysis of the controller's computational complexity has confirmed that the presented method can be successfully implemented in large-scale structures that are equipped with active control devices. Our follow-up research will validate the performance of the designed control on a real environment platform and we will design an adaptive controller to mitigate vibration in semi-active structures. • An adaptive algorithm for seismic active control is studied. • Online autoregressive model of the earthquake's signal is employed. • The method of alpha-shift allows for online solution to the optimization problem. • 20-story benchmark structure is employed to examine the performance of the control. • The controller outperforms the Linear-Quadratic-Gaussian regulator.