BACKGROUND: Rational medical decisions should be based on the best possible evidence. Clinical trial results, however, may not reflect conditions in actual practice. In hypertension, for example, trials indicate equivalent antihypertensive efficacy and safety for many medications, yet blood pressure frequently remains uncontrolled, perhaps owing to poor compliance. This paper examines the effect of initial choice of treatment on persistence with therapy in actual practice. METHODS: The authors examined all outpatient prescriptions for antihypertensive medications filled in Saskatchewan between 1989 and 1994 by over 22,000 patients with newly diagnosed hypertension whose initial treatment was with a diuretic, beta-blocker, calcium-channel blocker or angiotensin-converting-enzyme (ACE) inhibitor. Rates of persistence over the first year of treatment were compared. RESULTS: After 6 months, persistence with therapy was poor and differed according to the class of initial therapeutic agent: 80% for diuretics, 85% for beta-blockers, 86% for calcium-channel blockers and 89% for ACE inhibitors (p