Substitutive cell therapy using fetal striatal grafts has demonstrated preliminary clinical success in patients with Huntington's disease, but the logistics required for accessing fetal cells preclude its extension to the relevant population of patients. Human embryonic stem (hES) cells theoretically meet this challenge, because they can be expanded indefinitely and differentiated into any cell type. We have designed an in vitro protocol combining substrates, media, and cytokines to push hES cells along the neural lineage, up to postmitotic neurons expressing striatal markers. The therapeutic potential of such hES-derived cells was further substantiated by their in vivo differentiation into striatal neurons following xenotransplantation into adult rats. Our results open the way toward hES cell therapy for Huntington's disease. Long-term proliferation of human neural progenitors leads, however, to xenograft overgrowth in the rat brain, suggesting that the path to the clinic requires a way to switch them off after grafting.