The effects of cannabinoid ligands were studied in rats responding under a repeated acquisition procedure. Each session rats were required to learn a different three-response sequence; every third correct completion of the sequence resulted in the presentation of a food pellet. Errors produced a brief timeout but did not reset the chain. Neither injections of the centrally inactive cannabinoid, cannabidiol (3.2-100 mg/kg i.p.), nor the endogenous ligand, anandamide (0.01-18 mg/kg i.p.), affected rate or accuracy of responding. In contrast, delta9-tetrahydrocannabinol (3.2-18 mg/kg i.p.) and the long-acting analog of the endogenous ligand, R-methanandamide (1-18 mg/kg i.p.), produced dose-related increases in the total percentage of errors and decreases in the rate of responding. The brain cannabinoid receptor antagonist SR141716A (1-32 mg/ kg) did not affect either accuracy or rate of responding when administered alone. A low dose of SR141716A (1 mg/kg), which had no effect when administered alone, antagonized the disruptive effects of delta9-tetrahydrocannabinol and R-methanandamide on rate and accuracy of responding and produced an estimated 3-fold shift to the right in the dose-effect curves. However, administration of SR141716A did not alter the effects of morphine. These results suggest that cannabinoid agonists produce disruptions of learning in rats through stimulation of the cannabinoid receptor. The data further suggest that whereas cannabimimetic agents can disrupt learning, the anandaminergic system may not be tonically involved in learning.