OBJECTIVES: Marijuana use among high school seniors increased during most of the 1970s, decreased throughout the 1980s, and has been increasing again during the 1990s. Earlier analyses of the classes of 1976 through 1986 attributed the historic trends during that period to specific changes in views about marijuana. This study examined whether recent increases in marijuana use among seniors and among students in earlier grades reflect similar processes. METHODS: Multivariate regression analyses were conducted on data from large annual nationwide surveys of high school seniors from 1976 through 1996 (approximate n = 61,000) and 8th and 10th graders from 1991 through 1996 (n's = 87,911 and 82,475, respectively). RESULTS: Individual lifestyle factors (grades, truancy, religious commitment, evenings out for recreation) correlated substantially with marijuana use but did not explain the historic changes in marijuana use. Rather, decreases in perceived risk of harmfulness and in disapproval can account for the recent increases in all 3 grades and for earlier decreases among seniors. CONCLUSIONS: These findings indicate that perceived risks and disapproval are important determinants of marijuana use. Accordingly, prevention efforts should include realistic information about risks and consequences of marijuana use.