Abstract: To improve the angular resolution of sonar in limited carrier space, the frequency diversity based multiple- input multiple-output (MIMO) sonar is proposed. The MIMO sonar is composed of a receiving uniform linear array (ULA) of M-elements and a transmitting ULA of N-elements, which transmit narrowband signals with different center frequencies and the same envelope. In this paper, the signal model of dense frequency division MIMO sonar is established, also the direction and phase domain multiple signal classification (DPD-MUSIC) method, for direction of arrival (DOA) estimation is proposed. Taking dual-frequency MIMO sonar as an example, the performance of DPD-MUSIC algorithm for frequency division MIMO sonar is compared with that of the MUSIC algorithm for single-input multiple-output (SIMO) sonar. Numerical simulation results show that the frequency division MIMO sonar can obtain better angle resolution and DOA estimation accuracy than the SIMO sonar with equal number of receiving elements.