TNF inhibitors significantly reduce disease activity and morbidity in patients with RA, but there are few studies on the use of these drugs in actual clinical practice. OBJECTIVES: To evaluate treatment adherence with TNF inhibitors and methotrexate in patients with rheumatoid arthritis METHODS: We studied new drug starts with TNF inhibitors and methotrexate in RA patients enrolled in the California Medicaid program for the period 1999-2002. Drug discontinuation was defined if there was no prescription refill for 2 consecutive prescription periods (60 days for methotrexate and etanercept and 16 weeks for infliximab). Treatment persistence was assessed in terms of the number of days of continuous therapy. RESULTS: A total of 77,757 patients had a diagnosis of RA during 1999-2002. Of these, 8224 patients had a new drug start on methotrexate, and 2700 on TNF inhibitors (1251 for infliximab and 1449 for etanercept). Patients on TNF inhibitors were more likely to be female and non-white. During the study period (1999-2002), 82.3% of patients who started on methotrexate switched or discontinued their therapy, or added another drug (TNF inhibitor or leflunomide). Of the 2700 new drug starts on TNF inhibitors, 1649 patients (61.1%) discontinued or switched therapy (78.1% for etanercept, 41.3% for infliximab, p <0.001 compared to methotrexate). The time to discontinuation was significantly higher in patients on TNF inhibitors (460.7 + 10.4 days) compared to those on methotrexate (364.5 + 4.5 days). Cox proportional hazard model analysis showed that after adjusting for age, gender and ethnic origin, the treatment discontinuation was still statistically significantly different between the two groups (p <0.001). CONCLUSIONS: RA patients started on TNF inhibitors tend to stay significantly longer on therapy and have a lower rate of discontinuation compared to those started on methotrexate, perhaps indicating a better effectiveness/toxicity trade-off.