Aim. To develop and test the validity of a scale measuring attitudes towards smoking in current and former cigarette smokers. Design and participants. In a first mail survey, we collected qualitative data from 616 smokers. In a second mail survey, we collected quantitative data from 529 smokers and ex‐smokers. We conducted a 16‐month follow‐up survey among 93 participants in the second survey. Setting. Geneva, Switzerland, 1995–98. Findings. The study resulted in a three‐dimensional, 18‐item scale: the “Attitudes Towards Smoking Scale” (ATS‐18). The scale was validated with reference to criteria of content‐, construct and predictive validity. The three subscales measure perceptions of adverse effects of smoking (10 items), psychoactive benefits (four items) and pleasure of smoking (four items). Internal consistency coefficients (0.85, 0.88 and 0.81) and test–retest correlations were high (0.90, 0.75, 0.89, respectively). Differences in attitude scores between smokers in the pre‐contemplation and preparation stages of change were ‐ 0.83, 0.71 and 1.23 standard deviation units, respectively. A differential score (advantages minus disadvantages of smoking) predicted smoking cessation in baseline smokers and relapse in baseline ex‐smokers. Conclusion. ATS‐18 is a valid and reliable instrument which can be used in both research and clinical settings.