This study proposes a method to estimate the direction of arrival (DOA) for multiple sources by using microphone array under aliasing condition. Many conventional DOA estimation methods based on the sparseness of time-frequency (T-F) components of speech signals have to limit the sensor distance because of avoiding spatial aliasing due to the use of phase difference. Here we propose a 3-dimentional DOA or both azimuth and elevation angles estimation method capable to treat even in the presence of spatial aliasing. To cope with this problem, a reliable T-F cell selection and new bandwidth control in the kernel density estimation. In order to verify the effectiveness of the proposed method, experimental simulations have been performed. The simulation results show that the proposed DOA estimation gives accurate estimation even for spatial aliasing cases.