Patient adherence to their health care protocols is important to encourage the best health outcomes in rheumatoid arthritis (RA); however, little attention has been given to assessing the psychometric properties of adherence measures in this patient population. The purpose of this study was to evaluate the psychometric properties of three existing self-report measures of medication adherence in a sample of patients with RA--the compliance-questionnaire-rheumatology (CQR), the Medication Adherence Report Scale (MARS), and the medication adherence scale (MAS). A cross-sectional study of 108 clinic patients with rheumatoid arthritis was conducted to evaluate the reliability and validity of the measures. Cronbach's alpha was .77 for both the CQR and a modified version of the MARS. For the MAS, the Kuder-Richardson 20 reliability was .25. Although not strong, test-retest reliability was adequate for all measures. Factor analysis indicated that both the MARS and the CQR measure two factors. All three instruments were moderately correlated with each other, with correlations ranging between .48 and .56. Although these scales were significantly correlated, moderate correlations among the scales indicate that they may not measure the same aspects of adherence. Among the three adherence measures, the modified MARS demonstrated the best evidence of reliability and validity and ease of administration in this sample of persons with RA.