The present paper outlines the reasons for the current interest in work on septo-hippocampal grafts. It examines the role of cholinergic dysfunction in the memory deficits associated with Alzheimer's disease, the effects of hippocampal lesions on memory in infra-human animals, and the anatomy of the hippocampus. Methodological aspects of neural grafting are then examined, including the source, nature and site of the graft. A review of the tasks employed to determine functional recovery following septo-hippocampal grafts suggests that although recovery is evident its nature is unclear. An experiment is described which suggests that grafts from embryonic septum bring about recovery of working memory in rats. Different bases of the recovery of function are discussed, including the role of the graft in eliciting release of trophic factors from the host brain; the possibility that the graft acts by providing a pool of neurotransmitter; and finally that the graft may replace the damaged circuitry of the host. Some problems of the grafting procedure are outlined. It is concluded that grafting may provide a viable treatment technique in the absence of other forms of treatment for Alzheimer's disease.