To report on the accuracy of an in vivo dosimetry (IVD)-based source tracking (ST) method for high dose rate (HDR) prostate brachytherapy (BT). Methods: The ST was performed on a needle-by-needle basis. A least square fit of the expected to the measured dose rate was performed using the active dwell positions in the given needle. Two fitting parameters were used to determine the position of each needle relative to the IVD detector: radial (away or toward the detector) and longitudinal (along the axis of the treatment needle). The accuracy of the ST was assessed in a phantom where the geometries of five HDR prostate BT treatments previously treated at our clinic were reproduced. For each of the five treatment geometries, one irradiation was performed with the detector placed in the middle of the implant. Furthermore, four additional irradiations were performed for one of the geometries where the detector was retracted caudally in four steps of 10-15 mm and up to 12 mm inferior of the most inferior active dwell position, which represented the prostate apex. The time resolved dose measurements were retrieved at a rate of 20 Hz using a detector based on an Al 2 O 3 :C radioluminescence crystal, which was placed inside a standard BT needle. Individual calibrations of the detector were performed prior to each of the nine irradiations.Results: Source tracking could be applied in all needles across all nine irradiations. For irradiations with the detector located in the middle region of the implant (a total of 89 needles), the mean AE standard deviation (SD, k = 1) accuracy was −0.01 mm AE 0.38 mm and 0.30 mm AE 0.38 mm in the radial and longitudinal directions, respectively. Caudal retraction of the detector did not lead to reduced accuracy as long as the detector was located superior to the most inferior active dwell positions in all needles. However, reduced accuracy was observed for detector positions inferior to the most inferior active dwell positions which corresponded to detector positions in and inferior to the prostate apex region. Detector positions in the prostate apex and 12 mm inferior to the prostate resulted in mean AE SD (k = 1) ST accuracy of 0.7 mm AE 1 mm and 2.8 mm AE 1.6 mm, respectively, in radial direction, and −1.7 mm AE 1 mm and −2.1 mm AE 1.1 mm, respectively, in longitudinal direction. The largest deviations for the configurations with those detector positions were 2.6 and 5.4 mm, respectively, in the radial direction and −3.5 and −3.8 mm, respectively, in the longitudinal direction. Conclusion: This phantom study demonstrates that ST based on IVD during prostate BT is adequately accurate for clinical use. The ST yields submillimeter accuracy on needle positions as long as the IVD detector is positioned superior to at least one active dwell position in all needles. Locations of the detector inferior to the prostate apex result in decreased ST accuracy while detector locations in the apex region and above are advantageous for clinical applications.