A review of all morphine classical conditioning studies performed since Pavlov reveals significant shifts in methodological and conceptual thinking during the past 7 decades. These shifts have tended to obscure the fact that a great deal of basic data are still needed before morphine classical conditioning can be understood. These data are needed to help clarify the relationship between drug conditioning and drug addiction. Recent evidence, for example, suggests that once morphine autonomic conditional reactions are established, they are highly resistant to extinction. The inclusion of these types of data in any conceptualization of drug dependency would appear to be of help in the formulation of new clinical approaches to this problem.