Fetal neural grafts, rich in dopamine neurons, taken from the ventral mesencephalon and implanted into the dopamine-denervated striatum, can reinnervate the striatum, form synaptic contacts with host neurons, release dopamine and improve motor function. In animal models of Parkinson's disease, the improvement resulting from transplantation is dependent on the number of surviving grafted dopamine neurons and the density and extent of graft-derived reinnervation. The major unresolved scientific question at present is not whether neural grafting is better than established drug treatments but if survival and function of such grafts are at all possible in patients with Parkinson's disease. A more general problem is that if cell transplantation is to become clinically useful for a large number of Parkinsonian patients and also be applied in other neurological disorders, alternative sources of donor tissue must be found; several have been proposed, including adrenal medulla cells and sympathetic ganglia but perhaps the most exciting strategy is to implant cells that have been genetically engineered to synthesize and release L-dopa or dopamine.