Repetitive cerebral ischemia in gerbils produces delayed neuronal damage in the substantia nigra reticulata (SNr). This damage begins 4 to 5 days after the insult and is severe by day 7. The damage can be attenuated by GABA agonists. There is a prominent GABAergic striatal pathway to the SNr. Damage to this pathway leads to progressive loss of SNr neurons. This loss can be prevented by GABA agonists. We postulate that, ischemia-induced lack of GABAergic inhibitory input from the striatum to the SNr, may be responsible for this delayed neuronal damage. In the present experiment, we have measured striatal extracellular GABA concentrations with or without nipecotic acid, a GABA-reuptake inhibitor, in gerbils exposed to repetitive ischemia. GABA levels were measured on days 1, 3, 5, and 7 after the ischemic insult. Five control animals and a similar number of ischemic animals were monitored on each day. Extracellular fluid was collected using in vivo microdialysis and GABA levels were measured by electrochemical detection with HPLC. The extracellular striatal GABA levels were very low in the initial three specimens collected, both in the control and in the ischemic animals. However, addition of nipecotic acid resulted in an immediate increase of GABA in measurable range. In comparison to the controls, the increase in GABA on day 1 and 3 were significantly higher in animals with repetitive ischemia (two-way ANOVA with repeated measures). Subsequent measurements showed a gradual decrease in GABA levels when compared to controls. The increase in GABA with nipecotic acid was significantly lower on day 7 after the ischemic insults when compared to the controls. The increased GABA responsiveness immediately after the ischemic insults may reflect a protective effect against excitotoxicity. The subsequent decline in GABA levels after the insult may be secondary to progressive loss of striatal GABAergic neurons. This may contribute to the production of delayed neural damage in the SNr by a decrease in the inhibitory striatal input.