INTRODUCTION Wisconsin Card Sorting Test (WCST) performance is often compromised in Parkinson disease (PD). OBJECTIVE We aimed to investigate (a) the role of working memory (WM) and information processing speed (IPS) in the WCST performance of PD without dementia and (b) the profile of PD without dementia optimal WCST performance. METHODS 73 nondemented patients with PD were examined in WCST, WM (Digit Span Backwards [DSPANB] and Arithmetic subscale, Wechsler Adult Intelligence Scale [WAIS]), and IPS (Digit Symbol subscale-WAIS and Trail Making Test, part A [TMT-A]) and compared to a group of 48 healthy participants. The group with PD was further divided into 2 subgroups on the basis of their WCST performance, number of categories achieved (CAT):0-2 and CAT:3-6, and comparisons were performed. RESULTS The DSPANB correlated significantly and was the only neuropsychological variable that significantly contributed to the WCST perseverative errors (WPERRORS) performed by the PD group. Differences in the cognitive performance between CAT:3-6 and CAT:0-2 PD subgroups were no longer significant after controlling for age and general level of intelligence (full IQ [FIQ]). Age and FIQ significantly contributed to the WPERRORS performed by CAT:3-6 PD subgroup, while DSPANB was the only variable that significantly contributed to their WCST overall performance. CONCLUSIONS Our findings address the relationship between WM and inflexible behavior exhibited by PD without dementia, argue for the importance of lower age, higher education, and level of intelligence for the successful performance on WCST; yet, the differences in cognitive performance regardless of the duration of illness within our patients with PD provide support to cognitive reserve concept.