Improvements in MPTP-induced deficits were only found in subjects that received fetal substantia nigra transplants into the caudate nucleus. The MPTP-induced deficits were assessed using an object retrieval task that examined cognitive and subtle motor performance and by behavioral observation to determine the overall status of the subjects. Subjects that were also moderately or severely impaired by MPTP administration but that received inappropriate donor cells or implant sites (cerebellum to CN or SN to cortex) did not show any evidence of behavioral recovery. These subjects could not respond on the task in the months after grafting and were sacrificed, showing no improvements in parkinsonian signs or healthy behavior signs, up to 5-6 months after surgery. Grafting of SN cells into the striatum of non-MPTP lesioned subjects failed to modify normal behavior or induce abnormal behavior determined by our 2 behavioral assessment methods. In those monkeys that received the appropriate transplants, TH immunohistochemistry revealed that cells of the fetal substantia nigra grafted into the caudate nucleus survived and extended neurites into the host striatum. Indeed, grafted dopamine neurons were often associated with appreciable innervation of the caudate nucleus and appeared to be well incorporated into the host brain. In contrast, examination of the striatum of subjects in the inappropriate-graft group (e.g., cerebellar cells grafted into the caudate) showed no evidence of TH staining within the graft or host caudate nucleus. This indicated that there was no evidence of dopamine neurons present in the grafted tissue and that the mere presence of a fetal graft did not appear to induce sprouting in these MPTP-treated subjects. Although behavioral recovery occurred in only those monkeys that received appropriate transplants (fetal SN to host CN) and not in those that received inappropriate grafts (fetal cerebellum to CN or fetal SN to cortex), the CSF HVA levels did not distinguish those monkeys with improved parkinsonism from those that remained severely parkinsonian. The finding that in some SN-CN grafted subjects reported here, there was evidence of increased dopamine and lowered HVA/dopamine ratio in the vicinity of the SN grafts (cf. Elsworth et al., 1990b) is consistent with the hypothesis that graft-derived or graft-induced dopamine production is responsible for behavioral recovery. In addition, the finding that CSF HVA levels in non-MPTP lesioned subjects were unchanged by fetal SN grafts further indicates that CSF HVA levels may not be sufficiently sensitive to changes in central dopamine production to reflect release of dopamine from relatively small grafts that may, in lesioned subjects, modify behavior.(ABSTRACT TRUNCATED AT 400 WORDS)