Purpose To determine the tolerance limit (TL) and action limit (AL) of gamma passing rates (%GP) for volumetric‐modulated arc therapy (VMAT) patient‐specific quality assurance (PSQA) according to the American Association of Physicists in Medicine (AAPM) Task Group (TG)‐218 recommendations, and to comparatively evaluate the clinical relevance of 2D %GP and 3D %GP. Methods PSQA was performed for 100 head and neck (H&N) and 73 prostate cancer VMAT treatment plans. Measurements were acquired using a cylindrical water equivalent phantom, hollow in the center, allowing measurements with homogeneous or heterogeneous inserts. The LINAC‐delivered dose distributions were compared to those calculated from the treatment planning system through the gamma index. TL and AL were determined through the computation of two‐dimensional (2D) %GP using the recommended acceptance criteria. Dose–volume histograms were reconstructed from the measurements using a commercially available software to detect the dosimetric errors (%DE) between the compared dose distributions. Utilizing the estimated dose on the patient anatomy, structure‐specific %GP (3D %GP) were calculated. The 3D %GP were compared to the 2D %GP ones based on their correlation with the %DE. Each metric's sensitivity was determined through receiver operator characteristic analysis. Results TL and AL were in concordance with the universal ones, regarding the prostate cancer cases, but were lower for the H&N cases. Evaluation of %DE did not deem the plans unacceptable. The 2D %GP and the 3D %GP did not differ significantly regarding their correlation with %DE. For prostate plans, %GP sensitivity was higher than for H&N cases. Conclusions Determination of institutional‐specific TL and AL allows the monitoring of the PSQA procedure, yet for plans close to the limits, clinically relevant metrics should be used before they are deemed unacceptable for the process to be of higher sensitivity and efficiency.