PurposeTo establish the reliability and accuracy of a UNIQUE Linac in delivering RapidArc treatments and assess its long term stability. Materials and methodsUNIQUE performance was monitored and analyzed for a period of nearly two years. 2280 Dynalog files, related to 179 clinical RapidArc treatments were collected. Different tumor sites and dose scheduling were included, covering the full range of our treatment plans. Statistical distributions of MLC motion error, gantry rotation error and MU delivery error were evaluated. The stochastic and systematic nature of each error was investigated together with their variation in time. ResultsAll the delivery errors are found to be small and more stringent tolerances than those proposed by TG142 are suggested. Unlike MLC positional errors, where a linear relationship with leaf speed holds, other Volumetric Modulated Arc Therapy (VMAT) parameters reveal a random nature and, consequently, a reduced clinical relevance. MLC errors are linearly related only to leaf speed no matter the shape of the MLC apertures. Gantry rotation and MU delivery are as accurate as major competing Linacs. UNIQUE was found to be reliable and accurate throughout the investigation period, regardless of the specific tumor sites and fractionation schemes. ConclusionsThe accuracy of RapidArc treatments delivered with UNIQUE has been established. The stochastic nature of delivery errors is proven. Long term statistics of the delivery parameter errors do not show significant variations, confirming the reliability of the VMAT delivery system.