Purpose During respiratory‐gated volumetric‐modulated arc therapy (VMAT), the radiation beam is turned off each time the target exits the gating window. At the same time, the gantry slows, stops, and rewinds before the beam is turned back on. A quality assurance (QA) test was developed to detect inaccuracies in the gantry angle position between beam‐off and beam‐on events during respiratory‐gated VMAT. Methods Strips of Gafchromic™ EBT3 film were taped to the surface of a Capthan® 504 phantom mounted at isocenter. A homogeneous dose was delivered to the films through a 2 cm × 10 cm slit in the jaws using a respiratory‐gated VMAT arc without the multileaf collimator. A periodic breathing cycle was used. Errors in gated delivery ranging from 0.5 to 5° were simulated by delivering nongated arcs with the same field size with over‐ and underlapping sections of 0.5–5°. The simulated errors were used to define QA levels to analyze the gated delivery. Results The QA test was capable of detecting errors as small as 0.5°. The test was delivered to three Varian TrueBeam™ linacs, and no gantry angle inaccuracies greater than or equal to 0.5° were detected on any of the films. Conclusions A QA test capable of detecting gantry angle inaccuracies at beam‐off and subsequent beam‐on as small as 0.5° was developed and implemented for Varian TrueBeam™ linacs.