We sought to define quantitative electroencephalographic (EEG) measures as biomarkers of both early and late cognitive decline in Parkinson's disease (PD). PD subjects classified as cognitively normal (PD-CogNL), mild cognitive impairment (PD-MCI), and dementia (PD-D) were studied. Cognitive status and neuropsychological testing was correlated with background rhythm and frequency band EEG power across five frequency bands. We conclude that global EEG measures have potential use as biomarkers in the study of both early and late cognitive deterioration in PD, including for evaluating its treatment. PD-MCI has mean quantitative EEG characteristics that represent an intermediate electrophysiological state between PD-CogNL and PD-D.