Over the past three decades, neuroimaging studies-including structural, functional and molecular modalities-have provided invaluable insights into the mechanisms underlying Parkinson disease (PD). Observations from multimodal neuroimaging techniques have indicated changes in brain structure and metabolic activity, and an array of neurochemical changes that affect receptor sites and neurotransmitter systems. Characterization of the neurobiological alterations that lead to phenotypic heterogeneity in patients with PD has considerably aided the in vivo investigation of aetiology and pathophysiology, and the identification of novel targets for pharmacological or surgical treatments, including cell therapy. Although PD is now considered to be very complex, no neuroimaging modalities are specifically recommended for routine use in clinical practice. However, conventional MRI and dopamine transporter imaging are commonly used as adjuvant tools in the differential diagnosis between PD and nondegenerative causes of parkinsonism. First-line neuroimaging tools that could have an impact on patient prognosis and treatment strategies remain elusive. This Review discusses the lessons learnt from decades of neuroimaging research in PD, and the promising new approaches with potential applicability to clinical practice.