Rats were trained in a two-lever operant chamber to discriminate the presence versus the absence of a drug. During drug sessions, the training procedure involved reinforcing presses on lever 1 with saccharin-sweetened water but not reinforcing presses on lever 2. During nodrug sessions, only presses on lever 2 were reinforced. After this discrimination was learned, each rat was trained to discriminate presence versus absence of a second drug. All rats learned this second discrimination. Finally, the rats were tested to determine whether they could still discriminate the first drug, as well as other pharmacologically related compounds; most rats could. Training drugs were phenobarbital 30-36, nicotine 0.4, amphetamine 0.4, cyproheptadine 7, phencyclidine 4, cyclazocine 1.5, fentanyl 0.04, and scopolamine 0.2 mglkg. The results indicate that drug versus no-drug discrimination training does not disrupt discriminative control previously established with a different drug. When considered in combination with the results obtained during substitution tests conducted after drug-versus-no-drug training, the data suggest that, instead of discriminating drug cues versus no-drug cues, rats discriminate presence versus absence of particular drug cues.Drug discriminations (DDs) can be established in a training compartment that allows either of two different instrumental responses by reinforcing one response during sessions conducted in one drug condition and the second response during sessions in a second drug condition. When animals are trained to discriminate between two different drugs (Dl vs. D2), there is general agreement that the "stimulus effects" of 01 act as discriminative stimuli for the first response, and that 02 provides discriminative stimuli for the second response. However, when animals are trained to discriminate a drug (D) versus no drug (N), there is some disagreement concerning the nature of the stimuli (cues) that underlie DDs and the type of learning that takes place during OD training.One formulation asserts that animals discriminate between cues present during 0 sessions (drug cues) and those present during N sessions (no-drug or saline-induced cues), that is, that the D response becomes contingent on cues produced by the training drug and that the N response becomes contingent on specific cues that occur only in the N state. For example, Brown, Feldman, and Moore (1968) state that