Degeneration of dopamine (DA) neurons in the midbrain underlies the pathogenesis of Parkinson's disease (PD). Supplement of DA via L-DOPA alleviates motor symptoms but does not prevent the progressive loss of DA neurons. A large body of experimental studies, including those in nonhuman primates, demonstrates that transplantation of fetal mesencephalic tissues improves motor symptoms in animals, which culminated in open-label and double-blinded clinical trials of fetal tissue transplantation for PD1. Unfortunately, the outcomes are mixed, primarily due to the undefined and unstandardized donor tissues1,2. Generation of induced pluripotent stem cells enables standardized and autologous transplantation therapy for PD. However, its efficacy, especially in primates, remains unclear. Here we show that over a 2-year period without immunosuppression, PD monkeys receiving autologous, but not allogenic, transplantation exhibited recovery from motor and depressive signs. These behavioral improvements were accompanied by robust grafts with extensive DA neuron axon growth as well as strong DA activity in positron emission tomography (PET). Mathematical modeling reveals correlations between the number of surviving DA neurons with PET signal intensity and behavior recovery regardless autologous or allogeneic transplant, suggesting a predictive power of PET and motor behaviors for surviving DA neuron number.