Purpose: Monte Carlo study of radiation transmission around areas surrounding a PET room. Methods: An extended population of patients administered with 18F-FDG for PET-CT investigations was studied, collecting air kerma rate and gamma ray spectra measurements at a reference distance. An MC model of the diagnostic room was developed, including the scanner and walls with variable material and thickness. MC simulations were carried out with the widely used code GEANT4. Results: The model was validated by comparing simulated radiation dose values and gamma ray spectra produced by a volumetric source with experimental measurements; ambient doses in the surrounding areas were assessed for different combinations of wall materials and shielding and compared with analytical calculations, based on the AAPM Report 108.In the range 1.5-3.0 times of the product between the linear attenuation coefficient and thickness of an absorber (μ x), it was observed that the effectiveness of different combinations of shielding is roughly equivalent. An extensive tabulation of results is given in the text. Conclusions: The validation tests performed showed a satisfactory agreement between the simulated and expected results. The simulated dose rates incident on, and transmitted by the walls in our model of PET scanner room, are generally in good agreement with analytical estimates performed using the AAPM Publication No. 108 method. This provides an independent confirmation of AAPM's approach. Even in this specific field of application, GEANT4 proved to be a relevant and accurate tool for dosimetry estimates, shielding evaluation and for general radiation protection use. © 2014 Associazione Italiana di Fisica Medica.