Combining multiple-input multiple-output (MIMO) radar and inverse synthetic aperture radar (ISAR) techniques can reduce the number of antennas used and shorten the radar integrated time compared with the single-channel ISAR for the same cross-range resolution. In existing MIMO-ISAR processing, the echoes of different sensors are rearranged into an equivalent single-channel ISAR signal. A new method without echo rearrangement is proposed for two-dimensional (2-D) MIMO-ISAR imaging. A 2-D frequency estimation algorithm based on Unitary ESPRIT and projection transformation is used to obtain the spatial and Doppler frequencies of scatterers, and a high cross-range resolution can be achieved. The relationship between the two frequencies is exploited to resolve the ambiguity of spatial frequency. The analysis and simulation results show that, compared with the existing method, the proposed method can decrease the relative rotation angle (or integrated time) required for imaging. Thus, this method is more suitable for imaging targets with limited rotation or high maneuvering.