The study findings showed that the univariate model of impulsivity/sensation-seeking, friend's use of tobacco, and self-reported alcohol use was significant, but when interaction terms were added to the model, univariate predictors became insignificant, and the interaction model accounted for more variance in smoking than the univariate model. Further, when three risk variables were present, adolescents were about 40 times more likely to be current smokers than when these elevations were not present. One implication of these findings is that prevention programs made need to be changed to target complex risk situations. A second implication is that the current tobacco-focused preventions that focus on simple skills training (typically refusal skills) may be inadequate for complex and widely reported high-risk situations. The findings may also have implications for research/intervention for other adolescent health-risk behaviors. The study sample consisted of 210 (56.67 percent female) students recruited from 1 public high school in a low-middle socioeconomic area of Brisbane, Australia. The youth ranged in age from 14 to 16. They were surveyed during the final term of the 2004 school year. Impulsivity and sensation-seeking were measured with the Impulsive Sensation-seeking subscale of the Zuckerman-Kuhlman Personality Questionnaire. Alcohol and tobacco use were assessed with quantity/frequency items within an Activities Checklist that assessed a range of health-related behaviors. 2 tables and 29 references