To use moving target signatures in synthetic aperture radar (SAR) for classification and identification, it is desirable to obtain a well-focused image from the blurred moving target signature. While a substantial amount of research has been done on solving the constant chirp rate defocusing of a moving target, this often leaves significant blurring in the moving target signature of a real target. This work aims at improving previous work on autofocus, specifically considering higher order motion terms, including those due to range walk and small rotations. Additionally, we aim to reduce the reliance on bright well-separated point scatterers. We apply SAR autofocus methods normally used for full imagery to moving targets: minimum entropy and phased gradient autofocus, and include recent innovations that allow these methods to handle range walk. Finally, we add a novel range-varying autofocus stage. We use SAR imagery of a truthed moving pickup truck collected with the MIT Airborne Radar Testbed (ARTB) to compare autofocus algorithms.