Objective: To evaluate the cost-effectiveness of etoricoxib (90mg), celecoxib (200/400mg), and the non-selective NSAIDs naproxen (1000mg) and diclofenac (150mg) in the initial treatment of ankylosing spondylitis (AS) in the UK. Methods: A Bayesian cost-effectiveness model was developed to estimate the costs and benefits associated with initiating AS treatment with etoricoxib, celecoxib, diclofenac, or naproxen. Efficacy, safety and medical resource and cost data were obtained from the literature. With mixed treatment comparison meta-analysis the obtained efficacy estimates were synthesized. Treatment benefit and degree of disease activity, as reflected with BASFI and BASDAI scores, were related to quality adjusted life years (QALYs) and disability related costs. Other cost outcomes related to drug acquisition, gastrointestinal and cardiovascular safety were taken into consideration. Uncertainty in the source data was translated into uncertainty in cost-effectiveness estimates and therefore decision uncertainty. Results: There was more than 98% a probability that etoricoxib results in greater QALYs than the other interventions. Over a 30-year time horizon, etoricoxib is associated with about 0.5 more QALYs than the other interventions. At 2 years there is a 77% probability that etoricoxib shows the lowest cost. This increases to > 99% at 30 years. At 30 years etoricoxib is expected to save £19,460 relative to celecoxib (200/400mg) and £14,140 relative to naproxen and diclofenac. For a willingness-to-pay ceiling ratio of £20,000 per QALY there is a >97% probability that etoricoxib is the most-cost-effective treatment. Additional analysis with different assumptions, including celecoxib 200mg, and ignoring cost-offsets associated with AS disability, supported these findings. Conclusion: This economic evaluation demonstrated that etoricoxib is the most cost-effective NSAID treatment for AS patients in the UK.