Because health behavior is often concerned with dynamic constructs, a longitudinal approach to measurement is needed. The Longitudinal Guttman Simplex (LGS) is a measurement model developed especially for dynamic constructs exhibiting cumulative, unitary development measured longitudinally. Data from the Television Smoking Prevention Project, a large media-enhanced school-based smoking prevention study, were analyzed using the longitudinal Guttman scales. Subjects were seventh graders at the time of the first measurement. Other measurements were taken 3 months later and 1 and 2 years after the second measurement. Two data sets were created because of changes in the questionnaire. Data from seventh graders comprised Data Set 1 which contained items on smoking behavior and intentions asked at all four measurement times. Eighth graders provided data for Data Set 2 which contained two waves of data plus items on alcohol, marijuana, and smokeless tobacco use. A number of sets of items thought to form longitudinal Guttman scales were analyzed. The results demonstrated the importance of peer pressure in convincing adolescents to smoke. Other results suggest that experimentation with cigarettes and hard liquor occur at about the same time. The LGS scales could be used as dependent variables in multivariate analyses, and as a means of identifying adolescents who may be at high risk for drug abuse. The measurement scales used in the study are appended. (NRB)