Smoking is common among patients with depression. Which starts first? A long-term study by Murphy et al. (p. 1663) provides suggestive information. Representative samples of people living in eastern Canada were selected and interviewed in 1952, 1970, and 1992. The overall prevalence of depression remained approximately the same (5.2%-5.7%), but the total prevalence of cigarette smoking rose from 41.9% to 46.4% and then fell to 32.1%. Only with the decline in smoking in 1992 was there a significant relationship between smoking and depression. Subjects who became depressed for the first time were more likely to continue smoking, start, or not quit than were subjects who were never depressed. Depression appeared to start first; there was no association between prior smoking and subsequent depression. Depressed people may be more willing to take risks, and there may be a payback in the form of "self-medication," i.e., counteraction of unpleasant feelings.