To characterize medication use in a "well" very old population and relate the quantity and type of medication use to 10-year mortality. A longitudinal, 10-year, follow-up study involving 488 healthy, community-dwelling volunteers aged 75 to 85 years. The subjects underwent a detailed baseline evaluation and annual assessments that included comprehensive physical exams, laboratory and diagnostic tests, and related interviews. Prescription and nonprescription drug use was determined by self report, confirmation through hospital records, and reports by subjects and significant others. At study baseline, the mean number of prescription and nonprescription medications used was 2.3 and 1.5, respectively. Female subjects (n = 315), those older than 80 years, or those who reported themselves to be in fair or poor health on initial health self-report were found to show significantly increased use of prescription medications. The most commonly used classes of medications were cardiovascular drugs and analgesics. Subjects who were consuming a greater number of prescription and nonprescription medications did not have higher mortality rates. After correcting for differences in cardiovascular health status between users and nonusers, only digoxin approached significance as an independent predictor of death (P < .08). This study confirmed that medication use in an ambulatory, old old population is not excessive. The oldest subjects in the cohort consumed more medications than did the younger subjects. Women used more prescription drugs than men. Increased medication use was associated with worse ratings on health self report. Medication use alone, however, was not a predictor of 10-year mortality in this population. Questions are raised about the inappropriate prescription of digoxin in older subjects.