Parkinson's disease (PD) is a neurodegenerative disorder in which progressive impairment of movement is caused predominantly by the loss of nigrostriatal dopaminergic neurons; hence, grafting dopamine-producing cells into the striatum is a rational therapeutic possibility. In this review we describe our experience over the past several years with nonhuman primate and PD patient studies of autologous adrenal medulla-to-caudate grafting and compare it with the results obtained by other investigators. Long-term follow-up evaluations of 10 of our PD patients has demonstrated that the procedure produced a significant behavioral improvement (22% average improvement in parkinsonian disability score at 7-12 months and 16% at 16-26 months, postgrafting, p < .01) and had a clinically meaningful impact that can be difficult to measure. However, we and others have failed to replicate the dramatic improvement reported initially by Madrazo and his colleagues. Most United States investigators feel that the degree of improvement seen with adrenal medulla grafts alone does not warrant the potential risk of serious morbidity or mortality that was experienced in some clinical investigations. In our parallel nonhuman primate studies, employing the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine model of hemiparkinsonism, we have found that cografting of autologous adrenal medulla and sural nerve into the caudate produces a much more striking degree of functional improvement, far greater grafted cell survival, and a ''sprouting'' response of host striatal dopaminergic fibers. Most experimental evidence now supports a trophic model of graft-host tissue interaction, and ongoing experiments are pursuing the mechanism of action in greater detail.