Thirty-four items concerning marijuana use were administered three times to a cohort of Canadian high school students between Grades 11 and 13, in 1977 (n = 439), 1978 (n = 419), and 1979 (n = 167). These items were designed to measure four distinct constructs: attitude, perceived peer approval or disapproval, concern about risks, and symbolic protest against conventional society. A ... [Show full abstract] cross-sectional factor analysis of the Grade 13 data supported the hypothesized factor structure, as had earlier work with college students. However, analogous analyses on the Grade 11 and Grade 12 data contradicted the hypothesized factor structure. Most notably, separate factors for positive and negative attitude appeared in the Grade 11 and 12 analyses. The data were interpreted primarily in terms of maturational changes, largely through elimination of alternative interpretations by internal analyses and previous findings. Implications were discussed for adolescent cognitive-attitudinal development, attitude theory, and analysis of panel data.