Studies show that tobacco consumption, primarily cigarette smoking, is the leading preventable cause of morbidity and mortality in society. Tobacco is also considered a gateway drug and may predict later illicit drug use, especially for females. Studies have also shown that tobacco use often begins in adolescence and the earlier its onset the more severe nicotine dependence is likely to be. This purpose of this study was to identify potential predictors of smoking initiation in adolescents. The study hypothesized that greater psychiatric symptom severity would be associated with greater risk of smoking. Participants for the study were 59 adolescents (25 healthy controls and 34 adolescents diagnosed with ADHD or conduct disorder (CD)) between the ages of 12 and 14 and without a history of substance use (including nicotine use) who were part of an ongoing longitudinal study of adolescents at risk for substance use disorders. Initial assessment included four visits (1-3 hours long) to the Intramural Research Program of the National Institute on Drug Abuse, and subsequent follow-up visits (1 hour long) scheduled every 4 months after completion of the initial assessment. The sample was divided into four groups: health control participants (Control group), ADHD without mood disorder co-morbidity (ADHD group), conduct disorder without ADHD (CD group), and co-morbid ADHD with mood disorders (Mixed group). Participants were followed for an average of 15 months. Aggression, conduct problems, hyperactivity, impulsivity, inattention, anxiety/depression, social difficulties, and somatic complaints were assessed at study entry, and tested as predictors for smoking. Results of the study showed that at the last follow-up 69.5 percent of the adolescents had not smoked, 17 percent had experimented with cigarettes, and 13.5 percent had smoked regularly. Behavior symptoms, particularly aggression and hyperactivity, predicted subsequent smoking status. Adolescents who initiated smoking were, at baseline, more aggressive, more hyperactive, and tended to present more conduct problems than the adolescents who did not smoke. In addition, attention and learning problems tended to be more severe as a function of smoking frequency. Analysis also revealed an unexpected association between somatic complaints and subsequent smoking; impulsivity, mood and anxiety, and social problems did not predict smoking initiation. Adolescents with externalizing psychiatric disorders (ADHD, CD) did not initiate smoking any more than the control group. These findings suggest that a dimensional (symptom severity) approach may identify at-risk adolescents earlier and more effectively than a categorical (psychiatric diagnoses) approach alone. Study limitations are discussed. 5 tables and 50 references