Extracellular levels of GABA, derived from cell suspension transplants of embryonic day 14-15 rat striatal primordia implanted into the previously excitotoxically lesioned striatum, were measured using intracerebral microdialysis in halothane-anaesthetized rats. GABA overflow was monitored using loop type dialysis probes implanted into grafted, age-matched ibotenic acid-lesioned and intact striata, under baseline conditions and after different pharmacological manipulations. Basal and evoked GABA release, which was reduced by 58 and 96%, respectively, in the excitotoxin-lesioned striatum, was restored by the striatal grafts to levels close to or above those observed in normal striata. The graft-derived release of GABA was most likely of neuronal origin, since the K(+)-evoked (100 mM) GABA overflow was reduced by almost 80% when Ca++ was replaced by 20 mM Mg++ in the perfusion medium, and blockade of GABA uptake by nipecotic acid (0.5 mM), induced a greater than six-fold increase in GABA overflow. However, perfusion of the graft with 1 microM tetrodotoxin in combination with K+ (100 mM) resulted in little if any reduction in the K(+)-evoked overflow. Histological analysis demonstrated a dense tyrosine hydroxylase-positive fibre network in the grafts, which was removed after a 6-hydroxydopamine lesion of the ipsilateral nigrostriatal pathway. The dopamine denervating lesion resulted in an increased K(+)-evoked GABA overflow both in the intact (+76%) and the grafted striata (+181%), suggesting that the tonic dopaminergic inhibitory control of GABA release, seen in the intact striatum, is also present in the grafted striata. The glutamate analogue, kainic acid (1 mM added to the perfusion fluid), evoked a 60-74% increase in GABA overflow both in intact striata (with or without dopaminergic denervation) and in the striatal grafts. This effect seemed to be dependent on an intact corticostriatal projection, since knife-cut transections of the frontal cortex at the level of the forceps minor, abolished the response in both the intact and grafted striata. These results demonstrate that grafts of fetal striatal tissue implanted into the excitotoxically lesioned striatum restore striatal GABA overflow in a neuron-dependent manner, close to or above that seen in the normal intact striatum. Furthermore, the graft-derived GABA release appears to be under normal regulatory control from the host dopaminergic and glutamatergic systems. Since the GABAergic striatal output system is critical for the expression of striatum-related behaviours, it is proposed that the graft-induced behavioural recovery in the striatal lesion model, at least in part, may depend on the restoration of striatal GABAergic neurotransmission.