Molecular neurobiological studies have yielded enormous amounts of valuable information about neuronal response mechanisms and their adaptive changes. However, in relation to addiction this information is of limited value because almost every cell function appears to be involved. Thus it tells us only that neurons adapt to 'addictive drugs' as they do to all sorts of other functional disturbances. This information may be of limited help in the development of potential auxiliary agents for treatment of addiction. However, a reductionist approach which attempts to analyse addiction at ever finer levels of structure and function, is inherently incapable of explaining what causes these mechanisms to be brought into play in some cases and not in others, or by self-administration of a drug but not by passive exposure. There is abundant evidence that psychological, social, economic and specific situational factors play important roles in initiating addiction, in addition to genetic and other biological factors. Therefore, if we hope to be able to make predictions at any but a statistical level, or to develop effective means of prevention, it is necessary to devise appropriate integrative approaches to the study of addiction, rather than pursue an ever-finer reductive approach which leads steadily farther away from the complex interaction of drug, user, environment and specific situations that characterizes the problem in humans.