Disease models of osteoarthritis (OA) have shown that COX-2-selective nonsteroidal anti-inflammatory drugs (NSAIDs, coxibs) may have beneficial effects on cartilage. Clinical or epidemiological evidence for this potential association is scarce. Theobjective of this study was to evaluate the risk of hip or knee replacement in users of coxibs compared to nonselective NSAIDs. A population-based case-control study was conducted with the Dutch PHARMO Record Linkage System. Cases (n = 26,202) had a first replacement of the hip or knee after enrollment (2000-2009). Up to two controls (without hip or knee replacement) were matched by year of birth, gender, healthcare region, and calendar year. Using conditional logistic regression analysis, odds ratios (ORs) for hip or knee replacement were estimated by comparing long-term (>/=1 year) nonselective NSAID use with long-term coxib use. Analyses were statistically adjusted for disease and drug history. Long-term use of nonselective NSAIDs was not associatedwith a different risk of hip replacement (adjusted OR = 0.89, 95 % CI 0.65-1.22) or knee replacement (adjusted OR = 0.74, 95 % CI 0.49-1.11) as compared to long-term coxib use. Results were not different after stratification by gender, age, and cardiovascular or gastrointestinal disease. This study shows that long-term users of nonselective NSAIDs do not have a different risk of hip or knee replacement as compared to long-term coxib users. Therefore, our results do not support that patients with OA could benefit from using coxibs in order to slow progression of this disease.