The authors conducted a comparative test of the hypotheses that (a) stress is an etiological factor for smoking and (b) cigarette smoking causes increases in stress (A. C. Parrott, 1999). Participants were a sample of 1,364 adolescents, initially surveyed at mean age 12.4 years and followed at 3 yearly intervals. Measures of negative affect, negative life events, and cigarette smoking were obtained at all 4 assessments. Latent growth modeling showed negative affect was related to increase in smoking over time; there was no path from initial smoking to change in negative affect. Comparable results were found for negative life events, with no evidence for reverse causation. Results are discussed with respect to theoretical models of nicotine effects and implications for prevention.