AbstractAlthough self‐report instruments are currently considered a valuable tool for measuring adherence, due to their low cost and ease of implementation, there are still important factors that impact measurement accuracy, such as social desirability and memory bias. Thus, the Global Assessment of Medication Adherence Instrument (GEMA) was developed to provide an accurate measure of this construct. The aim of this study was to evaluate the properties of the measurement of the Global Evaluation of Medication Adherence Instrument (GEMA) among patients with chronic diseases. A methodological study was conducted in the public hospital of the state of São Paulo, Brazil. The adherence to anticoagulants as well as the international normalized ratio (INR) was assessed on 127 patients. Besides GEMA, two other instruments were used to assess adherence: the Morisky Medication Adherence Scale‐8 (MMAS‐8) and the Measurement of Adhesion to Treatments (MAT). The GEMA presented a satisfactory level of specificity (0.76) to identify adherents among those with a stable INR, low sensitivity (0.43) for the identification of non‐adherents among those with an unstable INR, and a Positive Predictive Value of 0.70. Positive and weak to moderate correlations were observed between the proportion of doses assessed with GEMA and the scores on the MMAS‐8 (r = .26 and r = .22, respectively) and the MAT (r = .22 and r = .30, respectively). The GEMA presented good practicality, acceptability, and evidence of specificity regarding the stability of the INR. The validity of the construct was partially supported by the relationship with self‐reported measures of adherence.