Recent thinking about tobacco dependence has been influenced largely by a focus on the pharmacological effects of nicotine. We advocate a return to earlier views of dependence as comprising both pharmacological and non-pharmacological aspects. Moreover, we suggest that it may be profitable to reformulate research on dependence in terms of a bioinformational process model. The basic tenets of such a model are outlined, as are the challenges in exploring the nature of dependence simultaneously across the cognitive, physiological and behavioral domains of function.