OBJECTIVE: To compare, in real-life settings, the retention rates of initial anti-tumor-necrosis factor (TNF) treatments (etanercept [ETN], adalimumab [ADA] and infliximab [IFX]) used as first-line biotherapy for axial spondyloarthritis (axSpA), and evaluate treatment switches to another anti-TNF inhibitor in the event of treatment failure. METHODS: We analyzed the medical records of all SpA patients (Assessment in Ankylosing Spondylitis International Working Group axial criteria) treated with ETN, IFX or ADA between 2001 and February 2015. Drug retention rates were calculated using the Kaplan-Meier method and compared by means of the Cox extended model. Sub-analyses were performed according to discontinuation reasons. RESULTS: Of the 249 SpA patients analyzed (135 radiographic cases, 114 non-radiographic), 102 received ETN, 62 ADA, and 85 IFX. In total, 103 discontinued treatment. The retention rates of IFX, ADA and ETN were 67%, 59% and 56% after 3 years; 62%, 42% and 47% after 5 years; 55%, 42% and 24% after 8 years; 53%, 42% and 12% after 10 years, respectively. In multivariate analyses, the predictive factors for retention were: low BASDAI score (hazard ratio [HR]: 1.02 [1.01-1.04]), high C-reactive protein levels (HR: 0.98 [0.97-0.99]), concomitant disease-modifying therapy (HR: 0.4 [0.21-0.75]), and radiographic SpA (HR: 1.5 [1.0-2.52]). In total, 61 patients switched to another anti-TNF therapy. No difference was observed among the three anti-TNF therapies regarding median retention duration, although the retention rate proved higher for treatment switches from one monoclonal antibody to another. CONCLUSION: The retention rate in SpA patients proved high, with retention for IFX superior to that of ETN.