Clinically relevant graphic approaches to the assessment of patients' blood pressure responses to antihypertensive agents are reviewed and extended to include both statistical components and the effects of treatment-related patient discontinuations. Regression line fits of change in diastolic blood pressure (DBP) versus baseline diastolic blood pressure (BDBP) are used for multicenter comparisons of placebo, guanabenz, methyldopa, clonidine, propranolol, hydrochlorothiazide, and the combination of guanabenz plus hydrochlorothiazide. Orientations of the fitted lines can be contrasted to certain "ideal" response profiles and are particularly useful for comparative purposes. An additional graphic approach was developed to show overall "therapeutic" response rate categories, two of which account for patients discontinuing treatment because of nonresponse or adverse effects. This approach assesses antihypertensive effectiveness, "adjusted" for the effects of treatment-related patient withdrawal, thus addressing criticisms often raised in the evaluation of such agents. Although these methods are generally useful for comparing agents used in chronic disease therapy, further work is needed to extend these procedures to incorporate safety assessments and overall benefit-risk considerations in the selection of treatment modalities.