Past research based primarily on cross-sectional data, has suggested a Guttman scale of substance use onset where marijuana use is preceded by alcohol use, and where sometimes tobacco is placed between alcohol and marijuana. Prospective longitudinal data is needed to determine whether the stages form a temporal sequence. A new statistical methodology, Longitudinal Guttman Scaling, was used to investigate the hypothesis that there exist temporally ordered stages of substance use. Data were obtained from four waves of panel data from the High School Anti-Smoking Project, which took place between 1981 and 1983 in Los Angeles, California. Subjects were high school sophomores at the start of the study and seniors at the end. Results indicated that there are at least two stage sequences of substance use onset: Tobacco Use Onset, composed primarily of tobacco use items; and Multiple Heavy Use, consisting of alcohol, tobacco, and marijuana items. Both sequences begin with a general experimentation phase where alcohol, tobacco, and marijuana are tried before heavier use of any substance is adopted. This replicates the findings of previous studies, but suggests that they may hold only for initial experimentation. Somewhat weaker evidence was found for a Moderate Alcohol Use sequence, beginning with beer, wine, and liquor followed by moderate alcohol use. (Author/NB)