While Dixon with a multi-peak-fat model provides superb fat suppression over large imaging volumes, motion can be a challenge in specific anatomies. This work evaluates a comprehensive approach to effectively reduce remaining motion artifacts in Dixon TSE and FFE sequences, using Dixon decorrelation, partial averaging, saturation of the physiological motion artifact sources and in-phase (IP) – out-phase (OP) combinations.To noticeably reduce motion artifact in Dixon TSE and FFE sequences in anatomies hampered by motion, by providing a comprehensive Dixon acquisition and reconstruction approach.In a segmented k-space acquisition of multi-shot Dixon TSE sequences, motion within a short time leads to ghosting in the image. In the case where Dixon echoes are correlated, these motion ghosts are equivalent in amplitude when compared to standard TSE sequences. In the case where Dixon echoes are decorrelated, the repetitive motion ghosts are varied and averaged out. Simulations (MATLAB, The MathWorks, Massachusetts, United States) and phantom experiments were performed to demonstrate artifact levels of correlated and decorrelated motion using this novel Dixon decorrelation approach. Furthermore, the combination of this decorrelation approach with partial averaging and variable density sampling [1], where the k-space center is acquired more times and is denser than the periphery of k-space, is explored. Saturation pulses parallel to the excited slices and variable refocusing flip angle sweeps [2] are examined to reduce the flow ghosting artifacts from inflowing blood. Finally, the effect of motion on the generation of IP and OP images and synthetic water and fat images from the complex water and fat images has also been studied. Simulation experiments and volunteer scans were used to investigate a modulus combination approach. All experiments were performed on Philips 1.5T and 3.0T (Best, The Netherlands) clinical scanners. A phantom study evaluated the proposed improvements in a controlled setting, and a volunteer study was performed on twenty subjects for dual echo Dixon FFE and TSE acquisition in various applications: abdomen, lumbar spine, breast, knee, and shoulder.Phantom and free breathing abdominal volunteer scans were acquired with conventional Dixon TSE (Fig. 1 - top), Dixon TSE decorrelation (Fig. 1 - middle), and Dixon TSE decorrelation in combination with partial averaging (Fig. 1 - bottom). To simulate motion in a controlled phantom setting, a tube (red circle) was removed after one shot during a pause of the scan. Note the motion ghosting is much less visible when Dixon decorrelation is used compared to the standard correlated approach and is even further reduced with partial averaging (Fig. 1 - red arrows). Results from the volunteer study are shown in Fig. 1 - on the right side - for the abdomen. Respiratory motion is effectively reduced with the Dixon decorrelation approach compared to the standard Dixon TSE and is almost completely removed with the combination of Dixon decorrelation and partial averaging. Fig. 2 - top demonstrates the effectiveness of the saturation pulses in Dixon TFE removing the flow ghosting artifacts from inflowing blood from the abdominal aorta. The effect of refocusing flip angle sweeps to suppress flow ghosting artifacts in Dixon TSE are shown in Fig. 2 - bottom in the knee application. Flow ghosts are nicely reduced using the TSE refocusing flip angle sweep in combination with Dixon decorrelation. Additionally, partial averaging with variable density sampling can be combined with the Dixon decorrelation approach to further reduce physiological ghosting artifacts. The effect of modulus combination of water and fat images to generate Dixon IP and OP images is shown in Fig. 3 on simulations and volunteer scans. Motion ghosts introduced by an incomplete breath hold in the Dixon FFE sequence is effectively reduced with modulus combination.The combination of improvements including Dixon decorrelation, partial averaging, saturation of the physiological motion artifact sources using saturation pulses and refocusing flip angle sweeps, and modulus synthetic Dixon image combinations, reduce the sensitivity of Dixon FFE and TSE scans to motion and subsequently improve the diagnostic quality of the scans.