Internal dose assessment can be derived from the measurement of retained activity in the whole body or in an organ at a given time. In radiation protection, this assessment, so-called in vivo measurement, is performed by an external measurement of the subject with germanium detectors (in most cases). Calibration of these detectors is ensured by anthropomorphic phantoms which, for technical reasons, can only provide rough representations of human. It is especially the case for the chest phantoms used in lung counting, subject of this paper. This leads to substantial corrections on calibration factors that are particularly crucial and delicate in low-energy in vivo measurements, resulting in important systematic errors. In order to improve calibration, former work based on numerical phantoms associated with Monte Carlo computing techniques has already proven its benefits. To go further, a Graphical User Interface called "OEDIPE", a French acronym for "tool for internal personalized dose assessment", has been developing at the IRSN internal dose assessment laboratory, simulating real measurements using person-specific computational phantoms in association with MCNP calculation code. The study presented here is dedicated to the implementation and validation of a real in vivo monitoring system (AREVA/COGEMA Marcoule, France) equipped with 4 high purity germanium (HPGe) detectors. After modeling the facility and measurement geometry using OEDIPE (design and positioning of the detectors. . .), validation with different configurations was carried out in two steps: first with point sources (different nuclides, different source-detector distances) and then with the Livermore calibration phantom (different overlay plates, lungs contaminated with 241Am and a mixture of actinides). The final goal is to approach a personalized numerical calibration of the facilities in order to improve dose assessment, as the use of physical phantoms for calibration induces large uncertainties. Such application could be an opening door on a better activity assessment in nuclear medicine, especially in personalized dosimetry in radioimmunotherapy. © 2005 IEEE.