An analysis of the conditioning of drug effects is presented that permits the prediction of the nature and direction of conditioned responses and accounts for drug-conditioning results within stimulus-substitution theory. It is argued that the failure to define and identify correctly the unconditioned stimulus effect and the unconditioned response effect of a drug constitutes the source of the controversies about the relation of the conditioned response to the observed drug effect. A model is provided for specifying the site of action of a drug with respect to feedback systems and regulatory mechanisms, and methods for identifying the site of action are discussed. The ideas presented are relevant to studies of conditioned drive, to drug tolerance and sensitization, and to conditioning studies that do not involve drugs.Repeated administration of drugs often results in the conditioning of physiological responses. These conditioned responses can be distinguished from other direct and indirect drug effects by the fact that under appropriate circumstances, they can be elicited without administering the drug. This form of classical conditioning is important because it may occur whenever drugs are chronically administered. The drug administration ritual may act as the conditioned stimulus and come to elicit a conditioned response. These conditioned responses have been postulated to play a role in drug tolerance and sensitization (Siegel, 1975b(Siegel, , 1977b, in drug abuse (Grabowski & O'Brien, 1981;Lynch, Stein, &Fertziger, 1976;Wilder, 1948 Wilder, , 1973b,and in behavioral medicine (Woods & Kulkosky, 1976). However, the reader who sets out to find an explicit role for conditioned drug effects in specific situations is faced with a mass of contradictory findings and confusing interpretations. The most common of these