The action of δ-guanidinovaleric acid (DGVA) on the activity of the central nervous system was investigated in the rat using electroencephalography (EEG). EEG using epidural electrodes revealed initiation of spike discharges within 5-10min on the side of DGVA (synthesized from O-methylisourea and δ-aminovaleric acid, according to Weiss and Krommer) application on pia matter of the sensorimotor cortex (0.3nmol). Thereafter, the frequency became 15-20 spike discharges/min. Spike discharges in the EEG of the opposite cerebral hemisphere were observed 5-10min after the onset of spike discharges. Spike discharges lasted until the end of recording after 2h. Spike discharges induced by DGVA were completely suppressed within 10min by the supplimentation of (3R)-(－)-4-amino-3-hydroxybutanoic acid (L-GABOB)(1mM) or GABA (10mM) on the pia matter. In rats given L-GABOB topically together with DGVA, spike discharges were not induced. While L-GABOB showed a remarkable suppressive effect on spike discharges induced by DGVA, supplimenting α-amino-DGVA, i.e., arginine (Arg), (100mM) showed no effect on the discharges. DGVA induced spike discharges by topical application after pre-application of Arg. Neither phenobarbital (PB) (20mg/kg, i.m.) nor diazepam (DZ) (10mg/kg, i.p.), applied 20min after the DGVA application, showed any suppressing effects on spike discharges. DGVA induced spike discharges by topical application 20min after the injection of PB or DZ. Two anticonvulsants known to correlate with the GABA-receptor had no effect on spike discharges induced by DGVA. Sodium valproate (200mg/kg, i.p.) and diphenylhydantoin (20mg/kg, i.p.) applied 20min after the DGVA application showed no suppressing effects on spike discharges induced by DGVA. These findings suggest that DGVB might act directly on the GABA-receptor and might be a specific GABA antagonist.