The detection of cognitive decline in Parkinson's disease (PD), at the Mild Cognitive Impairment (MCI) stage, has prognostic and treatment implications. The Movement Disorders Society (MDS) has recently published criteria and guidelines for the diagnosis of possible and probable PD-MCI. In the present study we assessed the ability of the Scales for Outcomes in Parkinson's disease-Cognition (SCOPA-Cog) to discriminate possible PD-MCI cases from patients with PD-dementia (PDD) and from cognitively intact PD subjects. Hundred-and-thirteen consecutive PD patients underwent the MMSE, the Dementia Rating Scale and an interview on independence in daily living, and were classified as cognitively intact (n = 49), or as possible PD-MCI (n = 33) or PDD (n = 31), according to MDS criteria. Logistic regression analysis was carried out with PD-MCI diagnosis (yes/no) as an outcome variable, and age, education and the SCOPA-Cog total score as covariates. Classification of cases according to the regression model was used for constructing Receiver Operating Characteristic (ROC) curves. Area Under the Curve (AUC) was 0.92 [95% CI 0.86-0.98], for the differential diagnosis between PD-MCI and cognitively normal patients, and 0.97 [95% CI 0.80-1.00], for the differential diagnosis between PD-MCI and PDD. Sensitivity and specificity were 90% and 73% for the PD-MCI versus no cognitive impairment differentiation, at the cutpoint ≥24, and 93% and 97% for the PD-MCI versus PDD discrimination, at the cutpoint ≥17. The SCOPA-Cog is a quick and psychometrically sound PD-specific scale. Our findings support its use for the screening of possible PD-MCI.