A variety of antihypertensive drugs have been introduced into clinical practice at excessively high dose. Examples include most thiazide diuretics, propranolol, oxprenolol, atenolol, methyldopa, hydralazine and captopril. These very high doses have usually resulted from studies in which doses have been increased at regular intervals until the desired antihypertensive effect has been achieved or until unacceptable adverse effects have resulted. Frequently the starting doses were too high and the intervals between dose adjustment too short. In many cases these large doses resulted in unnecessary adverse effects--the adverse biochemical effects of thiazide diuretics, nephrotic syndrome, taste disturbances and neutropenia with captopril, the lupus syndrome with hydralazine and the central nervous system effects of methyldopa. Parallel group design with single doses and sufficient statistical power to distinguish between the upper and lower ends of the antihypertensive dose-response relationship should replace the dose-escalating design.