We studied the evolution in choice of antihypertensive therapy in the elderly over a seven-year period and defined factors associated with such prescribing. To accomplish this, we performed a retrospective analysis of 8428 enrollees older than 65 years of age in the New Jersey Medicaid and Medicare programs newly begun on antihypertensive therapy between 1982 and 1988. Diuretics were the most common agents prescribed (51%), followed by calcium channel blockers (14%), beta-blockers (13%), central adrenergic antagonists (11%), and angiotensin-converting enzyme inhibitors (5%). However, the use of diuretics declined sharply during the latter part of the time interval studied. Using logistic regression modeling, we determined that the odds of a subject being started on a diuretic compared with any other antihypertensive drug decreased from 1.0 during the referent years 1982 to 1984 to 0.75 in 1985 to 1986 and to 0.41 in 1987 to 1988, after controlling for demographics, comorbidity, and hospital and physician visits (P < .001). The relative odds of diuretic use were significantly increased (P < .05) for the oldest subjects (odds ratio, 1.28 for age 85 and older versus ages 65 to 74 odds), women (odds ratio, 1.15), and blacks (odds ratio, 1.14). Despite the growing evidence during the study interval of the efficacy of diuretics in treating hypertension in the elderly, diuretic use diminished throughout this period in relation to other antihypertensive drugs. Subject characteristics (oldest subjects, women, and blacks) were important determinants of physicians' choice of antihypertensive therapy.