<h3>Objective:</h3> To establish the diagnostic accuracy of the Montreal Cognitive Assessment (MoCA) when screening externally validated cognition in Parkinson disease (PD), by comparison with a PD-focused test (Scales for Outcomes in Parkinson disease–Cognition [SCOPA-COG]) and the standardized Mini-Mental State Examination (S-MMSE) as benchmarks. <h3>Methods:</h3> A convenience sample of 114 patients with idiopathic PD and 47 healthy controls was examined in a movement disorders center. The 21 patients with dementia (PD-D) were diagnosed using Movement Disorders Society criteria, externally validated by detailed independent functional and neuropsychological tests. The 21 patients with mild cognitive impairment (PD-MCI) scored 1.5 SD or more below normative data in at least 2 measures in 1 of 4 cognitive domains. Other patients had normal cognition (PD-N). <h3>Results:</h3> Primary outcomes using receiver operating characteristic (ROC) curve analyses showed that all 3 mental status tests produced excellent discrimination of PD-D from patients without dementia (area under the curve [AUC], 87%–91%) and PD-MCI from PD-N patients (AUC, 78%–90%), but the MoCA was generally better suited across both assessments. The optimal MoCA screening cutoffs were &lt;21/30 for PD-D (sensitivity 81%; specificity 95%; negative predictive value [NPV] 92%) and &lt;26/30 for PD-MCI (sensitivity 90%; specificity 75%; NPV 95%). Further support that the MoCA is at least equivalent to the SCOPA-COG, and superior to the S-MMSE, came from the simultaneous classification of the 3 PD patient groups (volumes under a 3-dimensional ROC surface, chance = 17%: MoCA 79%, confidence interval [CI] 70%–89%; SCOPA-COG 74%, CI 62%–86%; MMSE-Sevens item 56%, CI 44%–68%; MMSE-World item 62%, CI 50%–73%). <h3>Conclusions:</h3> The MoCA is a suitably accurate, brief test when screening all levels of cognition in PD.