Ninety-six male Sprague-Dawley rats were trained in one of seven drug versus saline (SAL) discrimination (DD) tasks under a variable-ratio 5-15 schedule of food-motivated lever press responding. Three groups of rats (n = 12/group) were trained to discriminate between one of the legal over-the-counter (OTC) stimulants--caffeine (CAF), ephedrine (EPHED), phenylpropanolamine (PPA), and SAL. Three other groups (n = 12/group) were trained to discriminate between one of three binary stimulant combinations--CAF+EPHED, CAF+PPA, EPHED+PPA, and SAL. The seventh group of rats (n = 24) was trained to discriminate between SAL and a ternary combination of the OTC stimulants, CAF+EPHED+PPA. Generalization tests were conducted with each of the OTC stimulants and the controlled stimulants--amphetamine (AMPHET) and cocaine (COC). The data suggest: 1) there is cross-generalization between some OTC combinations and controlled stimulants; 2) full generalization between the OTC and controlled stimulants were demonstrated in rats trained to discriminate two of the binary stimulant combinations from SAL; 3) drug mixtures are not perceived as new entities distinct from their component elements; 4) training dose-ratio may influence the characteristics of mixture discriminations; 5) stimulus overshadowing may be a factor determining drug mixture cues, and 6) the DD properties of aggregate drug compounds may function within a euclidean metric space. We propose that some binary OTC stimulant combinations may effectively function as a methadone-like replacement therapy in cocaine dependence.