The GABA/benzodiazepine receptor complex in the basal ganglia of primates treated with the neurotoxin n-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) has been studied by semi-quantitative autoradiography with [3H]flunitrazepam ([3H]FNZ). Systemic treatment with MPTP produced a stable and lasting parkinsonian condition, with pronounced bradykinesia, akinesia and tremor. In the lateral segment of the globus pallidus (GPL) there was a significant reduction of [3H]FNZ binding compared with non-treated animals. There were no significant changes in the [3H]FNZ binding in the caudate nucleus, putamen and medial globus pallidus (GPM). This suggests that MPTP-treatment increases GABA release within the GPL exclusively. In view of the available evidence suggesting increased striatal output, and reduced unit activity within the GPL of the MPTP-treated primate, it seems likely that the striatal GABAergic output to the GPL is overactive in this model of Parkinson's disease. Furthermore, as there is no evidence for a change in GABA function within the GPM using this measure, the striatal neurones which innervate the GPM may be differentially affected by loss of dopamine innervation. In line with structural evidence and extrastriatal dopamine receptor distribution this suggests that the two striatopallidal systems are functionally heterogeneous. A hemi-parkinsonian primate model has also been used in this study. This model was produced by injection of MPTP directly into one carotid artery. The substantia nigra pars compacta (SNc) was destroyed on the injected side alone, and consequently the appearance of parkinsonian symptoms was confined to the contralateral side. [3H]FNZ binding in the GPL appears to be bilaterally reduced in this model, suggesting an interaction between the treated and non-treated side of the brain. In addition there is increased binding in the putamen and GPM with respect to the non-treated side of the brain. The increased [3H]FNZ binding in the GPM of the unilateral model may be due to the greater disruption of the nigropallidal and/or nigrostiatal dopamine neurones relative to the systemic model. The former would have the effect of uncoupling D1 dopamine receptors located on the terminals of striatal efferents from nigropallidal dopamine input, and as D1 dopamine receptors are implicated in the presynaptic control of GABA release from the terminals of striatal efferents, this would consequently reduce the level of GABA release in the GPM. The latter possibility would suggest that striatopallidal neurones projecting to GPM are more resistant to the effects of dopaminergic denervation than those projecting to GPL.