Objective: To determine the cost effectiveness of arthrocentesis as initial treatment compared to care as usual (CAU) for temporomandibular joint (TMJ) arthralgia. Materials and methods: 80 patients were randomly allocated to arthrocentesis as initial treatment (n = 40) or CAU (n = 40). Arthrocentesis was performed under local anaesthesia. The joint was rinsed with at least 300 ml isotonic saline chloride and no additional substances or drugs were applied. CAU consisted of a soft diet, physical therapy (which involved a home exercise program, joint mobilisation, and physical treatment modalities), and splint therapy using an intra-oral hard acrylic splint. Follow-up was after three, 12 and 26 weeks. Main outcome measures: Cost effectiveness, i.e. total costs from a societal point of view related to TMJ pain as measured on a visual analogue scale (VAS 0-100 mm) and cost utility, i.e. quality adjusted life years (QALY). Results: After 26 weeks, in both groups (arthrocentesis n = 36; CAU n = 36) TMJ pain had declined. Generalized estimating equation (GEE) models for the clinical effectiveness, based on intention to treat, showed significant differences between arthrocentesis as initial treatment and CAU (regression coefficient beta =-10.76 (95% confidence interval -17.75 to -3.77), p = 0.003). Estimated mean total (societal) costs during 26 weeks were D 589 ($795) in the arthrocentesis group and D 1680 ($2266) in the CAU group. With regard to cost effectiveness, arthrocentesis dominated CAU in about 98% of the bootstrap simulations.With regard to cost utility, arthrocentesis dominated CAU in approximately 95% of the bootstrap simulations. Conclusions: The results of this study suggest that arthrocentesis as initial treatment is cost effective in reducing TMJ pain compared to CAU. However, due to substantial missing data, current results with regard to cost effectiveness should be interpreted with caution.