Most patients with Parkinson disease (PD) develop both cognitive and motor impairment, and biomarkers for progression are urgently needed. Although α-synuclein is altered in cerebrospinal fluid of patients with PD, it is not known whether it predicts motor or cognitive deterioration. We examined clinical data and α-synuclein in >300 unmedicated patients with PD who participated in the deprenyl and tocopherol antioxidative therapy of parkinsonism (DATATOP) study, with up to 8 years of follow-up. Longitudinal measures of motor and cognitive function were studied before (phase 1) and during (phase 2) levodopa therapy; cerebrospinal fluid was collected at the beginning of each phase. Correlations and linear mixed models were used to assess α-synuclein association with disease severity and prediction of progression in the subsequent follow-up period. Despite decreasing α-synuclein (phase 1 to phase 2 change of −0.05 ± 0.21 log-transformed values, P < 0.001), no correlations were observed between α-synuclein and motor symptoms. Longitudinally, lower α-synuclein predicted better preservation of cognitive function by several measures [Selective Reminding Test total recall α-synuclein × time interaction effect coefficient, −0.12 (P = 0.037); delayed recall, −0.05 (P = 0.002); New Dot Test, −0.03 (P = 0.002)]. Thus, α-synuclein, although not clinically useful for motor progression, might predict cognitive decline, and future longitudinal studies should include this outcome for further validation.