Objective: Inflammatory back pain (IBP), the key symptom of axial spondyloarthritis (axSpA), including ankylosing spondylitis, has been proposed as a screening test for patients presenting with chronic back pain in primary care. The diagnostic accuracy of IBP in the rheumatology setting is unknown. Methods: Six rheumatology centres, representing secondary and tertiary rheumatology care, included routinely referred patients with consecutive chronic back pain with suspicion of axSpA. IBP (diagnostic test) was assessed in each centre by an independent (blinded) rheumatologist; a second (unblinded) rheumatologist made the diagnosis (axSpA or no-axSpA), which served as reference standard. Results: Of 461 routinely referred patients, 403 received a final diagnosis. IBP was present in 67.3%, and 44.6% (180/403) were diagnosed as axSpA. The sensitivity of IBP according to various definitions (global judgement, Calin, Berlin, Assessment of SpondyloArthritis international Society criteria for IBP) was 74.4%-81.1 % and comparable to published figures, whereas the specificity was unexpectedly low (25.1%-43.9%). The resulting positive likelihood ratios (LR+) were 1.1-1.4 and without major differences between sets of IBP criteria. The presence of IBP according to various definitions increased the probability of axSpA by 2.5%-8.4% only (from 44.6% to 47.1%-53.0%). Conclusions: The diagnostic utility of IBP in the rheumatology setting was smaller than expected. However, this was counterbalanced by a high prevalence of IBP among referred patients, demonstrating the effective usage of IBP in primary care as selection parameter for referral to rheumatology. Notably, this study illustrates potential shifts in specificity and LR+ of diagnostic tests if these tests are used to select patients for referral.