Stem cell transplantation offers a potentially transformative approach to treating neurodegenerative disorders. The safety of cellular therapies is established in multiple clinical trials, including our own in amyotrophic lateral sclerosis. To initiate similar trials in Alzheimer's disease, efficacious cell lines must be identified. Here, we completed a preclinical proof-of-concept study in the APP/PS1 murine model of Alzheimer's disease. Human neural stem cell transplantation targeted to the fimbria fornix significantly improved cognition in two hippocampal-dependent memory tasks at 4 and 16 weeks post-transplantation. While levels of synapse-related proteins and cholinergic neurons were unaffected, amyloid plaque load was significantly reduced in stem cell transplanted mice and associated with increased recruitment of activated microglia. In vitro, these same neural stem cells induced microglial activation and amyloid phagocytosis, suggesting an immunomodulatory capacity. Although long-term transplantation resulted in significant functional and pathological improvements in APP/PS1 mice, stem cells were not identified by immunohistochemistry or PCR at the study endpoint. These data suggest integration into native tissue or the idea that transient engraftment may be adequate for therapeutic efficacy, reducing the need for continued immunosuppression. Overall, our results support further preclinical development of human neural stem cells as a safe and effective therapy for Alzheimer's disease.