Purpose or Objective: While in-room Magnetic Resonance Imaging starts becoming part of radiotherapy (RT) treatments, the use of X-ray imaging equipment in Image-Guided RT (IGRT) is still growing and with it the need to evaluate the additional dose-to-organs it delivers. This study aims at verifying the accuracy of Monte Carlo (MC) calculation of the patient dose-to-organs delivered by four commercially available kV imaging systems: the XVI CBCT (Elekta), the OBI CBCT (Varian), the ExacTrac 2D-kV system (Brainlab) and the 2D-kV CyberKnife imaging system (Accuray). Simulations were validated against OSL measurements in the pediatric anthropomorphic phantom Grant (CIRS, ATOM) performed in three different clinical sites. Material and Methods: Each of the four kV-imaging systems was modeled as a Virtual Source Model (VSM) using the Penelope MC code. Such models were validated as part of a previous study using ionization chambers in water phantoms [G. Boissonnat et al., ESTRO 2017 Vienna]. In a second step, CT images of the phantom Grant were used to generate a voxelized phantom by converting the HU value of each voxel into the appropriate biological tissue (chemical composition and density). Then for each system, photons produced by the corresponding VSM were propagated in the voxelized phantom in order to obtain the 3D relative absorbed dose-to-medium map for three localizations (head, thorax and pelvis). MC-calculated doses were calibrated in amplitude using the ratio between the air kerma measured with an ionization chamber at the isocenter and the corresponding simulated value. After calibrating OSLs in air kerma at every beam quality, OSL measurements were performed in the anthropomorphic phantom at three localizations (head and neck, thorax and pelvis). After verifying that beam quality inside the phantom was impacting OSL corrections factors of less than 5%, they were neglecting. Therefore measured air kerma values were converted into absorbed dose-in-medium values using the incoming beam quality before being compared to simulated dose values. Results: MC calculations were performed in 2 hours on a cluster of 40 CPUs with a MC uncertainty better than 5% in 1mm3 voxels. The current study highlights the possibility to reproduce absolute dose measurements using VSM-driven MC simulations with an overall agreement better than 20 % (inside the irradiation field) for all four kV imaging systems and for the three anatomical localizations as presented in Table 1. Conclusion: This study demonstrates that MC calculations based on VSMs allow obtaining reliable absolute doses for kV imaging protocols in a reasonable computing time. All these developments are currently integrated into a dedicated software for imaging dose prediction, which will also include the Tomotherapy MVCT imaging system [V. Passal et al., MCMA 2017 Napoli].This software will enable to study the magnitude of additional doses delivered by in-room X-Ray imaging positioning units during the course of a complete RT treatment.