Measurement of patient compliance is essential if management of low compliance is to be performed efficiently. We assessed the value of several easily obtained clinical assessments compared to quantitative pill counts among 134 newly treated hypertensive male steelworkers during the first 6 months of their treatment with antihypertensive medication. Patient's self-reports obtained on structured interview correlated best with pill count compliance (r = 0.74, p less than 0.0001). Patients overestimated their compliance by an average of 17% but 90% of those who admitted to being noncompliant were found so. Qualitative urinary chlorthalidone and hydrochlorothiazide levels and changes in serum potassium, uric acid, and blood pressure also correlated with pill count compliance but were less accurate than interviews. Assessment of the patient's "health beliefs" and a variety of sociodemographic and health traits and perceptions did not provide useful information on compliance. Interviewing the patient is a simple and useful approach in assessing compliance with antihypertensive therapy.