The authors compared self-reported medical history and medication use in a cataract case-control study of 1,380 persons (1985-1989) in Boston, Massachusetts, with information from the participants' physicians. Under- and overreporting varied by condition and type of medication. A self-reported history of hypertension had the highest sensitivity (91%), and diabetes history had the highest specificity (97%). Among different medications investigated, self-reported antihypertensive medication use was the most sensitive (88%), while self-reported use of insulin was the most specific (99%). Differences between patient- and physician-reported frequencies were very small, except for arthritis (15%) and regular aspirin use (21%). Results suggest an accurate recall of medical and drug usage history in well-defined chronic conditions.