Do smokers smoke in order to meet attentional and/or emotional needs? This thesis examines natural smoking behaviour in cognitive tasks and during social interaction. Physiological, behavioural, subjective and social domains of behaviour were integrated and examined simultaneously across time. The thesis begins with an overview of key theories in a) smoking and b) emotion. Each of these is followed by a separate critical review of the current research on a) smoking and cognitive performance and b) emotion, respectively. These two sets of literature share methodological and theoretical problems: natural behaviour is commonly constricted to meet the demands of the experiment but not the subject. In a between-S's design, and a second study which examined subjects between treatments, the role of smoking during cognitive processing was examined. In eaach study nonsmokers (NS), smokers not smoking (SNS) and smokers smoking (SS) were observed while performing Raven's Matrices puzzles. Heart rate (HR), skin conductance response (SCR), smoking acts (puffs and taps) and performance (total number correct and reaction time) were measured. Although smokers are believed to perform better than nonsmokers and smokers not smoking (eg Wesnes & Warburton, 1983), such performance findings were not demonstrated in these studies. A number of other findings replicated in the two studies which may contribute to our understanding of why smokers smoke during a high level cognitive task: reliable HR and SCR response curves were found throughout the task; easy, medium and difficult slides were discriminated by the physiological curves; smoking acts were distributed nonrandomly throughout the task; and, immediate antecedent and consequent HR (for males only), SCR and body movement (BM) effects were found to surround puffs. The second study, which also included mood measures, found that males and females were affected differentially by smoking on physiological measures: females, but not males, had significantly reduced SCR, and males, but not females, had significantly more BM. These physiological effects probably serve to focus attention for smokers (Knott, 1984: O'Connor, 1985). The suggestion is that performance is not enhanced for smokers since elevated and sustained HR is not able to decrease sufficiently to perform with maximal efficiency (Lacey & Lacey, 1974). The final study examined smoking and emotional behaviour in 23 female mature students while discussing important aspects of their lives with an interviewer. Fourteen subjects were smokers. Two separate studies were first conducted in order to acquire a) a reliable code of face expression, and b) material which was meaningful to the subjects. HR, SCR, BM, paralinguistic and face exapression, speech, self-report and smoking acts were measured throughout the interview while personality questionnaires were administered after. This final study suggests that emotional behaviour must be described in terms of a multivariate model and that face expression may be an imortant variable to consider in future research on smoking in social situations.