The objectives of this review are, first, to summarize results from clinical trials on patients with Parkinson’s disease using grafts of either adrenal medulla or fetal substantia nigra and, second, to discuss some of the scientific issues that need to be clarified in more detail before transplantation can become a real therapeutic alternative in Parkinson’s disease. Adrenal medulla autotransplantation using open microsurgery has been shown to effect a modest reduction of the duration and severity of off periods in about 30–50% of operated patients and to have a high morbidity and mortality rate. The mechanisms of improvement are unknown. It is concluded that adrenal medulla autotransplantation is an experimental approach and not a treatment for Parkinson’s disease. Animal experimental data clearly favour the use of fetal dopamine (DA) neurons in patients, and clinical trials using such grafts are now going on in several countries. We have implanted human fetal mesencephalic tissue into the striatum in 6 patients. The findings indicate that fetal DA neurons can survive in the human parkinsonian brain and produce therapeutically valuable functional effects. Together with the solid animal experimental data the clinical observations support the idea that neural transplantation can be developed into an effective therapy in Parkinson’s disease. However, it should be emphasized that such a treatment is presently not available and that further work is necessary to optimize the transplantation procedure, e.g. with respect to the yield of surviving DA neurons and the location and number of implantation sites to achieve the largest possible symptomatic im