The most practical method of assessing medication compliance is to ask the patient for a self-report; but the interview question has to be selected and validated. The aim of this work was to select and validate interview questions to be used in a self-administered questionnaire for measurement of medication compliance in hypertensive patients. In 117 treated hypertensive patients, we compared an interview question to a clinical evaluation of medication compliance. Kappa indices were calculated to test the concordance between interview question and medication compliance. In 184 patients treated for hypertension since at least a year, the validated self-report was compared to a clinical evaluation of medication compliance. Interview questions were not equally sensitive for detection of medication compliance and 6 questions were included in the Compliance Evaluation Test (CET). In 184 patients, we compared the results obtained with CET to clinical evaluation of medication compliance. Values of kappa indices were above 0.5 in "good compliance" when "No" was answered to the 6 items, in "minor noncompliance" when 1 or 2 "Yes" were given, in "noncompliance" when 3 or more "Yes" were given. In this population of hypertensive patients followed in a hypertension clinic, we observed 10% "noncompliance", 24% "minor noncompliance", and 66% "good compliance". In clinical practice, interview may be the most useful method of measuring medication compliance. We demonstrated that the compliance evaluation test is validated and may help physicians to face the problem of nonadherence among their hypertensive patients.