OBJECTIVES: The aim of this study was to assess the predictive validity of four self-reported measures of adherence on glycemic control measured with HbA1c. METHODS: A survey conducted to assess factors associated with adherence to non-insulin antidiabetes drugs in the Canadian province of Quebec serves as the background for the present study. Participants completed an on-line questionnaire in which adherence to their treatment was assessed using four self-report instruments: the 4-item and 8-item Morisky Medication Adherence Scales (MMAS-4/8), an adaptation of a 5-item scale previously developed to be used with HIV patients and a 5-point Likert single item scale developed by our team. A sample of those who completed the questionnaire was then asked between 3 and 6 months later to measure their HbA1c.   We plotted a receiver operating characteristics (ROC) curve for each adherence measure and glycemic control (HbA1c ≤7% - >7%). The predictive performance of each instrument was assessed using the area under the ROC curve (AUC). AUC ranges from 0 to 1, with 0.5 indicating a no better than chance prediction.  RESULTS: A total of 117 participants were studied. Non-stratified analyses yielded an AUC of 0.515 (95% CI: 0.423-0.606) for the MMAS-4, 0.532 (0.431-0.633) for the MMAS-8, 0.541 (0.452-0.629) for the HIV-adapted scale, and 0.524 (0.441-0.607) for our scale. CONCLUSIONS: Overall all 4 self-reported measures of adherence exhibited a poor validity at predicting glycemic control.