To develop effective smoking prevention interventions, we need to identify modifiable variables, such as the "pros and cons" of smoking, that predict self-initiated smoking cessation. Our objective was to assess associations between the pros and cons of smoking and the stages of change, both cross-sectionally and longitudinally, in a cohort of smokers. In cross-sectional comparisons, the pros of smoking were 0.19-0.31 standard deviation (SD) higher and the cons 0.79-0.87 SD lower in the precontemplation than in the preparation stage. In follow-up data, progressing from precontemplation to contemplation and from contemplation to preparation was associated with substantial and significant increases in the cons (+ 0.71 SD and + 0.50 SD, respectively). No longitudinal associations were found between changes in pros of smoking and progress through stages. Thus this study added evidence from longitudinal data to published evidence from cross-sectional data about the association between the cons of smoking and the stages of change. Intervention studies are necessary to establish whether this association is causal. (C) 1999 Elsevier Science Ltd.