The effect of three anticonvulsants on several test methods for evaluation of memory-enhancing drugs have been studied in rats and mice. The results were compared to the results obtained from the nootropic piracetam and from imipramine. In an active avoidance test (pole jumping), repeated administration of carbamazepine (5 mg/kg i.p.) and piracetam (300 mg/kg p.o.) protected against impairment of learning rate caused by repeated application of an electroconvulsive shock. Valproate calcium (repeated administration of 30 mg/kg i.p.) was only weakly active, while clonazepam (repeated administration of 0.3 mg/kg i.p.) decreased the learning rate even more. Drugs were given in dosages which have no anticonvulsive activity in this test. The impairment of learning rate caused by repeated application of ethanol was prevented by carbamazepine (5 mg/kg i.p.), valproate calcium (30 and 100 mg/kg i.p.) and piracetam (100 mg/kg i.p.). Clonazepam (0.3 and 1 mg/kg i.p.) had a worsening effect on learning rate. In Porsolt's behavioral despair test, carbamazepine (5, 10 and 30 mg/kg i.p.) and valproate calcium (30 and 100 mg/kg i.p.) shortened the duration of immobility, which indicates an increased psychomotor activity. Clonazepam was ineffective. Results obtained with carbamazepine are similar to those obtained with piracetam. The data are discussed in view of the so-called psychotropic effects of carbamazepine in clinical trials.