Objective: The goal of this study was to compare the microRNA (miRNA) profile of Parkinson's disease (PD) frontal cortex with normal control brain, allowing for the identification of PD specific signatures as well as study the disease-related phenotypes of onset age and dementia. Methods: Small RNA sequence analysis was performed from prefrontal cortex for 29 PD samples and 33 control samples. After sample QC, normalization and batch correction, linear regression was employed to identify miRNAs altered in PD, and a PD classifier was developed using weighted voting class prediction. The relationship of miRNA levels to onset age and PD with dementia (PDD) was also characterized in case-only analyses. Results: One twenty five miRNAs were differentially expressed in PD at a genome-wide level of significance (FDR q < 0.05). A set of 29 miRNAs classified PD from non-diseased brain (93.9% specificity, 96.6% sensitivity). The majority of differentially expressed miRNAs (105/125) showed an ordinal relationship from control, to PD without dementia (PDN), to PDD. Among PD brains, 36 miRNAs classified PDD from PDN (sensitivity = 81.2%, specificity = 88.9%). Among differentially expressed miRNAs, miR-10b-5p had a positive association with onset age (q = 4.7e-2). Conclusions: Based on cortical miRNA levels, PD brains were accurately classified from non-diseased brains. Additionally, the PDD miRNA profile exhibited a more severe pattern of alteration among those differentially expressed in PD. To evaluate the clinical utility of miRNAs as potential clinical biomarkers, further characterization and testing of brain-related miRNA alterations in peripheral biofluids is warranted.