Objectives - This study examined whether controlling for educational background in the CERAD cognitive screening battery would affect the likelihood of patients with Parkinson's disease to fulfill criteria for mild cognitive impairment (PD-MCI) and dementia (PDD). Materials & Methods - One-hundred seventeen patients with PD were studied. Cognitive impairment was determined as two subtest scores falling below either the standard cutoff scores or education-corrected cutoff scores. The presence of dementia was determined by clinical interview or Clinical Dementia Rating. Patients were then classified as PD-MCI and PDD according to cognitive test performance and presence/absence of dementia. Results - The number of cognitively impaired patients (PD-MCI or PDD) was significantly higher when education-controlled cutoff scores were used (62.5% vs 38%). Correspondingly, the number of false negatives (demented PD patients performing normally in CERAD) was significantly lower when education-corrected cutoff scores were used (4% vs 10%). Conclusions Controlling for education increases the sensitivity of the CERAD for PD-MCI and PDD.