OBJECTIVES: Establish the Compliance Questionnaire Rheumatology (CQR) potential value as a predictive adherence tool by identifying similarities, differences and overall relationship between the CQR and the Morisky Medication Adherence Scale (MMAS-4). METHODS: Patients residing in the United States completed a self-administered, Internet-based questionnaire in the fall of 2011.  Patients self-reported a diagnosis of rheumatoid arthritis (RA).  The cross sectional survey included the CQR, MMAS-4 and extensive treatment and demographic patient level data.  CQR predicts patients that are likely to be adherent at 50%, 60%, 75%, 80%, 85%, 90% or 95% levels.  The MMAS-4 is scored from 0-4, with zero equal to perfect adherence.  Frequency distributions were compared.  Linear models looked at: inter-item correlations, CQR score thresholds optimally differentiating adherent vs. non-adherent on MMAS-4, and ordinary least squares (OLS) analysis of the ability of the CQR to predict MMAS-4 scores. RESULTS: Survey respondents were 76.2% female, 86.2% Caucasian, with mean age 56.4 years.  Frequency distributions of the CQR and MMAS-4 were similar. The CQR provided much more detail at high adherence levels, and hence appears more discriminative in these patients.  Comparing dichotomous adherent/non-adherent results, the optimal CQR threshold for predicting “adherent patients” from MMAS-4, was 60%.  Correlation between MMAS-4 and CQR was 0.40.  CQR scores can predict MMAS-4 scores, although the relationship appears weak. CONCLUSIONS: Compared with MMAS-4, the CQR scale appears to be more sensitive at high levels of adherence.  The CQR appears to be useful as a predictive tool. It does not require claims-based data to assess historical non-adherence, and so may be a useful alternative.  These results warrant further exploration of the CQR as a way to stream rheumatology patients into appropriate treatment, based on their potential to be non-adherent.