OBJECTIVES: The purpose of this study is to compare the direct cost parameters on rheumatoid arthritis (RA) by using two studies in Turkey conducted via patient database record system and via expert opinion methodology in 2010. METHODS: The data included in this study was collected upon the opinions of 22 rheumatologists from 53 clinics in Turkey. The other study was completed by Turkish Health Insurance Funds Database (THIFD) --1920 patients selected for a 6-month period. RESULTS: Based on the patient database, 5% of the patients were hospitalized and 42% had outpatient visit. According to expert opinion, 7.2% of the patients were hospitalized and 77.12% had outpatient visit. Patient records state that non-COX inhibitors (78%) and immunosuppressants (71%), and disease-modifying anti-rheumatic drugs (DMARDs) (11%) of the patients were prescribed and biologic DMARDs are not specifically stated. Based on the rheumatologists, non-COX inhibitors (34.52%), immunosuppressants (39.76%), methotrexate (80.86%), sulfasalazine (31.76%) and biologic DMARDs (17.15%) were prescribed. Total annual cost (EUR€2,386) results of the patient database system comprised of pharmacy costs (EUR€1,747), outpatient costs (EUR€360), and inpatient costs (EUR€252). Annual cost from expert opinion was EUR€2917.03 with outpatient & inpatient costs (EUR€206.82)  and pharmacy and prostheses/orthoses cost (EUR€2710.21). CONCLUSIONS: Expert panel methodology’s important way to observe the real world practice. It’s observed that expert panel and patient database results are in line with each other. Due to the distribution of the pharmaceuticals, a difference of only EUR€531.03 was found. It can be observed that biologic DMARDs treatments are perceived to be prescribed more, however according to patient data, prescribed ratio of biologic DMARDs is relatively low. THIFD is used to extract patient costs where patient records are used for controlling reimbursement criteria, so some of the data may be misleading. Accurate performing of coding is key issue to minimize possible bias in use of patient data.