Purpose: To develop an in vivo dosimeter system for stereotactic body radiation therapy (SBRT) that can perform accurate and precise real-time measurements, using a microsized amount of a photostimulable phosphor (PSP), BaFBr:Eu 2+. Methods: The sensitive volume of the PSP was 1.26 9 10 À5 cm 3. The dosimeter system was designed to apply photostimulation to the PSP after the decay of noise signals, in synchronization with the photon beam pulse of a linear accelerator (LINAC), to eliminate the noise signals completely using a time separation technique. The noise signals included stem signals, and radioluminescence signals generated by the PSP. In addition, the dosimeter system was built on a storage-type dosimeter that could read out a signal after an arbitrary preset number of photon beam pulses were incident. First, the noise and photostimulated luminescence (PSL) signal decay times were measured. Subsequently, we confirmed that the PSL signals could be exclusively read out within the photon beam pulse interval. Finally, using a water phantom, the basic characteristics of the dosimeter system were demonstrated under SBRT conditions, and the feasibility for clinical application was investigated. The reproducibility, dose linearity, dose-rate dependence, temperature dependence, and angular dependence were evaluated. The feasibility was confirmed by measurements at various dose gradients and using a representative treatment plan for a metastatic liver tumor. A clinical plan was created with a two-arc beam volumetric modulated arc therapy using a 10 MV flattening filter-free photon beam. For the water phantom measurements, the clinical plan was compiled into a plan with a fixed gantry angle of 0°. To evaluate the energy dependence during SBRT, the percent depth dose (PDD) was measured and compared with those calculated via Monte Carlo (MC) simulations. Results: All the PSL signals could be read out while eliminating the noise signals within the minimum pulse interval of the LINAC. Stable real-time measurements could be performed with a time resolution of 56 ms (i.e., number of pulses = 20). The dose linearity was good in the dose range of 0.01-100 Gy. The measurements agreed within 1% at dose rates of 40-2400 cGy/min. The temperature and angular dependence were also acceptable since these dependencies had only a negligible effect on the measurements in SBRT. At a dose gradient of 2.21 Gy/mm, the measured dose agreed with that calculated using a treatment planning system (TPS) within the measurement uncertainties due to the probe position. For measurements using a representative treatment plan, the measured dose