In inverse synthetic aperture radar (ISAR) imaging of a moving target with complex motions, such as a maneuvering airplane and a fluctuating ship with oceanic waves, the time-varying Doppler phases induced by target motions will severely deteriorate the target imaging performance. In this paper, a novel ISAR imaging algorithm for moving targets with complex motions based on cubic phase signal (CPS) estimation is proposed. In the proposed method, the received signal echoes in a range cell are modeled as multicomponent CPSs after migration compensation, and then a scaled symmetric instantaneous autocorrelation function is constructed to remove the coupling relationship between time variable and lag-time variable caused by the second-order parameter. After removing the residual defocusing effect caused by the third-order parameter, a CPS can be well accumulated as a pole peak. In the meantime, these three parameters can be simultaneously estimated according to the peak position, and then a high-quality ISAR image can be obtained. The effectiveness of the proposed algorithm is demonstrated by the ISAR imaging results of simulated data.