Plastic scintillation dosimeters (PSDs) have favourable characteristics for small and composite field dosimetry in radiosurgery, however, imperfect corrections for the Cerenkov radiation contamination could limit their accuracy for complex deliveries. In this work, we characterize the dose and dose-rate linearity, directional dependence, and compare output factors with other stereotactic detectors for a new commercially available PSD (Exradin W1). We provide some preliminary comparisons of planned and measured dose for composite fields delivered clinically by a Cyberknife radiosurgery system. The W1 detector shows good linearity with dose (<0.5%) and dose rate (<0.8%) relative to the signal obtained using an ion chamber under the same conditions. A maximum difference of 2% was observed depending on the detector's angular orientation. Output factors for all detectors agree within a range of ±3.2% and ±1.5% for the 5 and 7.5 mm collimators, respectively, provided Monte-Carlo corrections for detector effects are applied to diode and ion chambers (without corrections the range is ±5.5% and ±3.1% for these two collimators). For clinical beam deliveries using 5 and 7.5 mm collimators, four of the six patients showed better agreement with planned dose for the PSD detector compared to a micro ion chamber. Two of the six patients investigated, however, showed 5% differences between PSD and planned dose, film measurements and the ratio of PSD and micro ion chamber signal suggest that further investigation is warranted for these plans. The W1 detector is a promising tool for stereotactic plan verification under the challenging dosimetric conditions of stereotactic radiosurgery.