On the basis ofthe visual conditional discrimination paradigm, a procedure has been developed that forces rats to retrieve propositional rules from reference memory (e.g., flashing light, go left; constant light, go right), Using operant boxes, rats were trained to perform a conditional visual (successive) discrimination task (go/go) on an FR10 schedule of reinforcement. During the first trial of a day's test session, incorreet responding did not result in the termination of the trial. The ratio between correct and incorrect leverpresses until completion of the FRlO at the correet lever on this first trial was considered a measure ofthe rat's ability to retrieve the correct rule. During the subsequent trials of a test session, incorrect responses led to the termination of the trial. Behavioral measures from this stage of a day's session were considered as an indication ofbehavioral performance. Scopolamine (0.1, 0.39, or 1.56 mg/kg) did not affect retrieval but impaired all measures of performance. Chlordiazepoxide (3.0, 5.0, or 10.0 mg/kg) similarly spared retrieval abilities but disrupted performance at the highest dose. Baclofen (1.56, 3.13, or 4.69 mg/kg) nonspecifically disrupted responding at the highest dose. Amphetamine (0.3, 0.56, or 1.0 mg/kg) generally reduced the number of correct responses. Thus, none ofthese treatments specifically interfered with the retrieval of information about well-learned stimulus-response rules. It is hypothesized that the retrieval of these rules is comparable with the automatie and implicit reactivation of nondeclarative information in humans, or even with priming. Furthermore, it seems likely that this process shows multiple neuronal representations and, therefore, remains widely insensitive to manipulations in individual neurotransmitter systems.