In this experiment, the observing-response procedure was adapted for use with drug self-administration. Rats' responding for oral ethanol was sometimes reinforced on a random-ratio schedule, whereas at other times it had no effect (i.e., extinction). Behavior producing stimuli associated with the otherwise unsignaled random-ratio and extinction periods (i.e., observing behavior) was acquired and maintained. In a vehicle control condition, both self-administration and observing behavior decreased, but observing decreased less rapidly proportionally to baseline than vehicle consumption. Thus, conditioned reinforcers may have persistent effects that are relatively independent of the current status of the primary reinforcer. The procedure allows long-term study of drug-associated conditioned reinforcement and provides independent indexes of the conditioned reinforcing and discriminative stimulus effects of drug stimuli.