© 2014 by the Society of Nuclear Medicine and Molecular Imaging, Inc. For phantom studies involving 90Y PET/CT, homogeneous solutions of 90Y, for example, 90Y citrate, are commonly used. However, the microsphere biodistribution of a postradioembolization liver is never homogeneous; therefore, such phantoms are physiologically unrealistic for simulating clinical scenarios. The aim of this work was to develop a safe and practical phantom capable of simulating the heterogeneous microsphere biodistribution of a postradioembolization liver. Methods: Gelatin (5%) was used to suspend 90Y resin microspheres, poured into plastic containers to simulate a liver with 2 tumors. Microspheres were added while the gelatin was maintained in a liquid state on a hot plate and continuously stirred with magnetic stir bars. The liquid microsphere mixture was then rapidly cooled in an ice bath while being stirred, resulting in a heterogeneous suspension of microspheres. The completed phantom was serially scanned by 90Y PET/CT over 2 wk. Results: All scans demonstrated a heterogeneous microsphere distribution throughout the liver and tumor inserts. Serendipitously, magnetic stir bars left inside the phantom produced CT artifacts similar to those caused by embolization coils, whereas pockets of air trapped within the gelatin during its preparation mimicked gas within hollow viscus. The microspheres and tumor inserts remained fixed and suspended within the gelatin, with no evidence of breakdown or leakage. Conclusion: A gelatin phantom realistically simulating the physiologic microsphere biodistribution of a postradioembolization liver is feasible to construct in a radiopharmacy.