We tested a theoretical model of individual differences in smoking cessation using a social-cognitive conception of the self-concept. We developed and validated measures of the smoker self-concept and the abstainer self-concept. Each scale was shown to have good internal reliability and construct validity and was distinct from other important predictive measures used in smoking research (e.g. Fagerstrom Tolerance Questionnaire, smoking rate, motivation, self-efficacy). Importantly, we demonstrated the predictive validity of the self-concept scales. The interaction of baseline measures of the smoker self-concept and abstainer self-concept predicted smoking status three months after treatment; subjects were most likely to be abstinent if they began treatment with a strong abstainer selfconcept and a weak smoker self-concept. This interaction held over and above baseline smoking rate, Fagerstrom Tolerance scores, and measures of motivation and self-efficacy to quit. The utility of social-cognitive individual difference models and potential patient-treatment matching interventions are discussed.