There seems to be little doubt that smoking can be dangerous to your health. A growing body of scientific data has linked the excessive use of tobacco to an increased incidence of a wide range of disorders including cancer of the lung, bladder and oral cavity, cardiovascular disorders such as coronary heart disease and atherosclerosis and pulmonary disorders such as bronchitis and emphysema (USPHS, 1975(USPHS, , 1877). yet, the development of effective and durable treatments for smoking behavior has been an elusive goal. While there are some encouraging signs of progress, recent literature reviews (Bernstein & McAlister, 1976; Frederiksen & Simon, in press; Lichtenstein & Danaher, 1976) continue to conclude that the treatment of smoking is characterized by good initial success but very high rates of "relapse". Despite our best efforts we are unable to effectively help anywhere from 20-80% of smokers attempting to quit. While the abovementioned reviewers agree that we do have far to go, they also argue that a "behavioral" approach offers us our best chance for advancement, both from a methodological and treatment outcome perspective.The purpose of the current paper is to present an approach to the analysis and modification of smoking risk called controlled smoking. This approach attempts to reduce the health risks associated with smoking by either promoting abstinence or smoking in a manner that minimizes the associated health risks (i.e., "safer" smoking). It is this last emphasis, safer smoking, that has differentiated controlled smoking from many other approaches. The investigation of this treatment alternative has forced a reassessment of smoking behavior and its control. This reassessment has implications for our models of smoking behavior, measurement strategies and approaches to risk modification. The remainder of this paper will examine the nature of smoking risk, present a model of smoking behaviour, and the implication for measurement strategies and risk modification.