Aim. To examine the distinctness of the DSM-IV substance abuse and dependence constructs in a large, general adolescent population. Design. Data were collected using the 1995 Minnesota Student Survey. Survey items were designed to correspond to DSM-IV diagnostic criteria for substance abuse and dependence. Settings. Public schools, alternative schools and area learning centers. Participants. Of the 78 800 students between the ages of 14 and 18 years who completed the survey, 18 803 reported substance use and at least one substance use disorder diagnostic criterion during the previous 12 months and were used for the analyses. The sample was divided randomly into two groups in order to conduct data analyses on one group (n = 9490) and confirm the findings in the other group (n = 9313). Measurements. Confirmatory factor analyses were conducted to test three competing factor structure models consisting of a single factor model, a two-factor model of distinct dimensions and a two-factor model with interrelated dimensions. Findings. The single factor and correlated two-factor models had similar parameter estimates and fit the data better than the competing two-factor model with distinct dimensions. Findings were confirmed in a second sample. Conclusions. The study findings indicate that DSM-IV substance abuse and dependence criteria may be more optimally structured as a unidimensional construct rather than as bidimensional constructs for adolescents.