We have implemented a scheme for simulating realistic dynamic PET data from real MR acquisitions. This toolkit uses a series of MR acquisitions, image registrations and segmentations. PET images are simulated assigning typical values to the segmented images, and manually inserting additional lesions. The data are simulated using analytic forward-projections (including attenuation) with STIR [1], providing a fast and simple simulation toolkit that can be used to efficiently study the performance of different reconstruction and motion and attenuation correction approaches for dynamic PET data.