Introduction Freezing of gait (FOG) is one of the most common walking problems in Parkinson's disease (PD). Impaired cognitive function is believed to play an important role in developing and aggravating FOG in PD. But some evidence suggests that motor function discrepancy may affect testing results. Therefore, we think it is necessary for PD-FOG(+) and PD-FOG(-) patients to complete neuropsychological tests under similar motor conditions. Methods This study recruited 44 idiopathic PD patients [PD-FOG(+) n = 22, PD-FOG(-) n = 22] and 20 age-matched healthy controls (HC). PD-FOG(+) and PD-FOG(-) patients were matched for age, year of education, and Hoehn and Yahr score (H&Y). All participants underwent a comprehensive battery of neuropsychological assessment, and demographical and clinical information was also collected. Results PD patients showed poorer cognitive function, higher risks of depression and anxiety, and more neuropsychiatric symptoms compared with HC. When controlling for age, years of education, and H&Y, there were no statistical differences in cognitive function between PD-FOG(+) and PD-FOG(-) patients. But PD-FOG(+) patients had worse motor and non-motor symptoms than PD-FOG(-) patients. PD patients whose motor symptoms initiated with rigidity and initiated unilaterally were more likely to experience FOG. Conclusion Traditional neuropsychological testing may not be sensitive enough to detect cognitive impairment in PD. Motor symptoms initiated with rigidity and initiated unilaterally might be an important predictor of FOG.