Background: Reliable estimation of radiation-absorbed dose is necessary to evaluate the benefits and the risks of radiopharmaceuticals used for diagnostic or therapeutic purposes in nuclear medicine. Materiel and Methods: This study included 47 patients treated with iodine-131 for Graves’ disease. A comparative study between Geant4 Monte Carlo simulation and MIRD formalism was carried out to evaluate the dose received by each patient. Patients’ thyroids and internal radiation were modeled using Geant4. Geant4 simulations were compared to experimental measurements performed with TLDs placed inside an ellipsoidal Thyroid phantom. MIRD was used to determine the beta doses received by the different patients. Results: The average difference between MIRD and Geant4 considering only beta emitted radiation was approximately 5.6%; this difference is justified by the fact that, Geant4, contrary to MIRD, considers all particle energies of the ¹³¹I spectrum, the shape of the thyroid and the heterogeneity of the dose deposited in the modeled volume. A good agreement was found between experiment and Geant4 simulations. The total dose received by patients varies between 176Gy and 359Gy. After 9 month, 74% of treated patients were rendered hypothyroid. Conclusion: This study showed the necessity of determining the specific activity of each patient considering the thyroid volume and the iodine fixation. It also revealed that the Geant4 toolis appropriate for accurate internal dosimetry calculations, particularly for the case of Graves’ disease treatment. GEANT4 can be used as a standard for the comparison of experimental measurements.