Beam-space transformation projects the array data into a lower space, which is not only effective in reducing computation time, improving performance, but also being capable to suppress interference. In contrast to conventional adaptive beam-space transformation method, which often requires adjusting the beam-space matrix and steering vectors online, an efficient adaptive beam-space transformation method is proposed. In the proposed method, the beam-space covariance matrix and the steering vector both have closed-forms, and do not depend on the adaptive beam-space matrix. This eliminates the online adjustment process, and, thus, improves the computational efficiency. Finally, the proposed method can also be applied to the direction of arrival (DOA) estimation. Simulation results demonstrate that it has a better DOA estimation performance than the conventional adaptive method. Furthermore, the proposed method also has another significant advantage, i.e., it is able to suppress moving interference. This can be ascribed to the proposed beam-space matrix which is independent of the historical data, and, thus, effective to avoid the mismatch between the training and application data, since this mismatch often occurs in conventional adaptive methods.