Neurocognitive disorders in Parkinson's disease (PD) are common and heterogeneous. The aim of this study was to use a data-driven method to describe different cognitive phenotypes in PD and to explore anxiety, depression, and motor disturbances across the different cognitive profiles. Latent profile analysis was applied to the neuropsychological performances of 65 patients with idiopathic PD assessed by means of a battery of tests that encompass measures of attention, memory, executive functions, social cognition, language, and visuospatial abilities. A three-cluster model produced the best solution: Cluster A (21.54%) included patients with intact cognition or with a relatively slight cognitive impairment in memory and executive functioning; Cluster B (53.85%) included patients with an intermediate level of cognitive impairment; and Cluster C (24.61%) included patients with the most severe cognitive impairment, with greater deficit compared to Cluster B in executive functioning, and, notably, in tasks with a predominantly posterior cortical basis (naming and visuospatial abilities). The three subgroups did not differ in terms of age, gender, disease duration, motor symptom severity or side of onset, levodopa equivalent daily dose, level of anxiety, or depression; however, patients from Cluster C showed greater impairment than patients from Cluster A in measures of everyday functioning. We presented a qualitative description of three distinct cognitive phenotypes emerging from a sample of 65 PD patients. The three clusters seem to be related to daily functioning but are independent from the stage of disease, motor functioning, anxiety, and depression. (PsycInfo Database Record (c) 2021 APA, all rights reserved).