Objective: To compare the costs of treating osteoarthritis (OA) pain using combination tramadol/paracetamol tablets, NSAIDs alone, NSAIDs plus proton pump inhibitors (PPIs), or NSAIDs plus histamine H2-receptor antagonists (H2RAs) from the perspective of the Dutch healthcare system. Design and methods: A decision-analytical model was constructed to model the cost outcomes of the four treatment strategies over 6 months. A cost-minimisation approach was used, which considered data related to resource utilisation, medication costs and costs for the treatment of adverse events. Data, derived mainly from the clinical literature, were supplemented by inputs from a Delphi panel as well as official price and tariff lists. The base-case analysis considered direct medical costs, including those for treating all adverse events with tramadol/paracetamol and gastrointestinal (GI) adverse events with NSAIDs. Separate scenario analyses explored costs of NSAID-based regimens: (i) according to 21 levels of risk for GI adverse events, and (ii) when renal events attributable to NSAIDs were considered. Robustness of the model was tested using univariate sensitivity analysis. Results: In the base-case analysis, costs for 6 months' treatment of OA pain using tramadol/paracetamol were €244.45, compared with €317.32 for NSAIDs + PPIs, €200.67 for NSAIDs + H 2RAs, and €125.86 for NSAIDs alone. This provided a cost saving of €72.87 per patient over 6 months for tramadol/paracetamol compared with NSAIDs + PPIs. Tramadol/paracetamol became cost saving compared with NSAIDs alone and NSAIDs + H2RAs for GI risk levels >13 and >10, respectively. When renal adverse events of NSAIDs were considered, tramadol/paracetamol was cost saving compared with all NSAID-based regimens (saving €228.40 vs NSAIDs, €418.42 vs NSAIDs + PPIs, and €302.69 vs NSAIDs + H2RAs [year of costing 2005]). Sensitivity analysis confirmed the model was robust to wide-ranging changes in the value of input parameters. Conclusion: Tramadol/paracetamol is cost saving ...