The cognitive status of patients with Parkinson's disease (PD) who developed pathological gambling (PG) during dopamine replacement therapy has been poorly explored. We compared clinical and cognitive features of 21 consecutive PD patients with active PG (PD-PG) versus 42 PD controls of similar disease duration without any impulse control disorder. All patients underwent full neuropsychological testing to evaluate executive and other frontal lobe-related functions, attention, learning and memory, language, visuospatial abilities and neuropsychiatric status [using Geriatric Depression Scale (GDS) and Neuropsychiatric Inventory (NPI)] as well as the South Oaks Gambling Screen Scale (SOGS). PD-PG were younger (60.4 vs. 64.9, p = 0.01) and more frequently of male gender (85 vs. 57%, p = 0.02). The two groups did not differ in medication dosages and kind of dopamine agonist. PD-PG had higher MMSE (29.1 vs. 27.4, p = 0.02) and performed better at Rey Auditory Verbal learning Test (45.9 vs. 40.4, p = 0.04), verbal phonemic fluencies (38.7 vs. 31.8, p = 0.02), verbal semantic fluencies (44.9 vs. 37.4, p = 0.01) and attentive matrices (47.6 vs. 43.5, p = 0.05) while the remaining cognitive performances were comparable to controls. Moreover, based on the NPI, PD-PG had higher aggressiveness, irritability, disinhibition and eating disorders than controls. In conclusion the occurrence of PG in our cohort of patients with PD was associated with preserved executive functions.