Objectives. Primary inefficacy of TNF inhibitors (TNFi) for axial spondyloarthritis (axSpA) is infrequent. The objective of this study was to assess the long-term evolution and final diagnosis of patients with primary inefficacy of TNFi for axSpA. Methods. This was a systematic retrospective study of all patients receiving a TNFi for axSpA in one tertiary referral centre. Patients had axSpA confirmed by a rheumatologist and were started on a first course of TNFi according to usual practice. If the rheumatologist interrupted treatment at 3 months for inefficacy, this was defined as primary inefficacy. Five to 10 years later, these patients were re-evaluated. Results. Of 222 patients receiving a first TNFi for axSpA, 27 (12%) were considered as having primary inefficacy. These patients were more often females (48 vs 27%, P = 0.04), had higher functional impairment [BASDAI (0–100) 68 vs 42, P = 0.001] and less increased CRP (50 vs 78%, P = 0.008.) At the follow-up, 25 (92%) patients were re-evaluated: the diagnosis of axSpA was confirmed for 21/25 (84%) patients according to the Assessment of SpondyloArthritis criteria and 20/25 (80%) patients according to the rheumatologist; but 18/25 (72%) had at least one other cause of their symptoms from among OA, widespread pain syndrome or depression. A second TNFi was prescribed for 16 patients and was efficacious for 9 (56%). Conclusion. Most patients with primary inefficacy had a confirmed diagnosis of axSpA, but they often had other causes of pain. We suggest that patients with primary inefficacy to TNFi should be screened for comorbidities that may interfere with axSpA activity assessment.