In renal dynamic contrast enhanced (DCE) MRI respiratory motion of the kidneys necessitates registration of the dynamics. Since image contrast varies during contrast agent passage, automatic registration is challenging. We show that on Dixon-derived fat-images this contrast change is virtually absent. Therefore, we propose to perform automated image registration using fat-images and apply the resulting transformation to the water-images. We applied this method to DCE data of 10 patients and show its superiority over a conventional registration approach. Pharmacokinetic fits to a two-compartment model yielded realistic values for renal perfusion and filtration.