Parkinsonism has been reported to improve following transplantation of fetal mesencephalic tissue into the striatum of MPTP-exposed monkeys and humans and in patients with idiopathic Parkinson's disease. While there is good evidence for the survival of grafted tyrosine hydroxylase (TH)-positive cells in animal studies, it is not known whether they produce neuronal effects that account for behavioral improvement after transplantation or whether spontaneous or graft-induced changes in the host striatum are at least partly responsible. Are neuronal synaptic connections and dopamine release necessary, or would "toenails and talcum powder" do the job equally well? We have addressed these questions by studying several types of implantation surgeries, including sham surgery, the implantation of cerebellar tissue, and the implantation of mesencephalic TH-positive fetal tissue of various gestational ages into the striatum. Adult male African green monkeys received systemic MPTP administration (cumulative doses of 2.0-2.5 mg/kg) prior to these stereotaxic surgical manipulations. Subjects were matched for quantitative behavioral deficits prior to surgery. Subjects were examined and assessments made by "blinded" observers who scored individual spontaneous and elicited behaviors. Observers were trained and tested repeatedly for inter-rater reliability. A "parkinsonian summary score" derived and determined using a principal component factor analysis of a large sample of data from MPTP-treated and normal monkeys of the same species was used to assess behavior. Postmortem brain tissue was prepared for biochemical analysis of dopamine concentrations and TH immunohistochemical studies. The most dramatic improvement was seen in monkeys with "early" (< 4 cm fetal crown rump length) surviving substantia nigra grafts in the caudate nucleus. Some behavioral improvements were seen in MPTP-treated sham-operated monkeys, cerebellar-grafted monkeys, and "later" (> 14 cm fetal crown rump length) substantia nigra-grafted monkeys. These changes in monkeys which did not have surviving dopamine-producing grafts probably represent the recovery capacity of MPTP-treated host brain during this time interval since un-operated subjects showed similar changes. More variable effects were seen with substantia nigra grafts in the putamen. The most consistent correlate of behavioral improvement in all experimental groups was elevation in dopamine concentrations near the grafts compared with a distant striatal location which is believed to represent the depletion without the effects of the grafts. While these data do not establish the precise mechanism of action, they point to a hierarchy of factors which provide increasingly larger restorative effects, including sprouting of host neurons and increased dopamine production by grafted fetal dopamine neurons. Sham surgery appears to be significantly less effective than early fetal mesencephalic tissue which survives and releases dopamine.