Objectives. To evaluate the accuracy of PET/CT and of PET vascular activity score (PETVAS) in assessing disease activity and the ability of PETVAS in predicting relapses in a large single-centre cohort of patients with large vessel vasculitis (LVV).Methods. We conducted a retrospective cohort study of prospectively collected data of consecutive patients diagnosed with LVV who underwent at least one PET/CT scan between 2007 and 2020. The nuclear medicine physician's interpretation of each PET/CT scan (active/inactive vasculitis) was compared with disease activity clinical judgement (active disease/remission). For each PET/CT scan, the PETVAS score was calculated and its accuracy in assessing disease activity was evaluated. The ability of PETVAS in predicting subsequent relapses was evaluated.Results. A total of 100 consecutive LVV patients (51 large vessel GCA, 49 Takayasu arteritis) underwent a total of 476 PET/CT scans over a mean follow-up period of 97.5 months. Physician-determined PET/CT grading was able to distinguish between clinically active and inactive LVV with a sensitivity of 60% (95% CI 50.9, 68.7) and specificity of 80.1% (95% CI 75.5, 84.1); the area under the curve (AUC )was 0.70 (95% CI 0.65, 0.75). PETVAS was associated with disease activity, with an age and sex-adjusted odds ratio for active disease of 1.15 (95% CI 1.11, 1.19). A PETVAS 10 provided 60.8% sensitivity and 80.6% specificity in differentiating between clinically active and inactive LVV; the AUC was 0.73 (95% CI 0.68, 0.79). PETVAS was not associated with subsequent relapses, with an age and sex-adjusted hazard ratio of 1.04 (95% CI 0.97, 1.11).Conclusions. The visual PET/CT grading scale and PETVAS had moderate accuracy to distinguish active LVV from remission. PETVAS did not predict disease relapses.