Background: The objective of this study was to validate disease-related metabolic brain patterns for Parkinson’s disease, multiple system atrophy, and progressive supranuclear palsy. Methods: The study included 20 patients with Parkinson’s disease, 21 with multiple system atrophy, and 17 with progressive supranuclear palsy, all of whom had undergone a clinically motivated [18F]-fluoro-deoxyglucose positron emission tomography scan at an early stage of their disease. At a follow-up time after the scan of 2–4 years, a clinical diagnosis was made according to established clinical research criteria. Patient groups were compared with 18 healthy controls using a multivariate covariance image analysis technique called scaled subprofile model/principal component analysis. Results: Disease-related metabolic brain patterns for these parkinsonian disorders were identified. Validation showed that these patterns were highly discriminative of the 3 disorders. Conclusions: Early diagnosis of parkinsonian disorders is feasible when the expression of disease-related metabolic brain patterns is quantified at a single-subject level.