Background. The Functional Gait Assessment (FGA) is a validated measurement of gait-related activities in certain populations and may be potentially useful to assess balance and gait disorders in patients with Parkinson disease (PD). Objective. The purpose of this study was to determine the construct, concurrent, and predictive validity of the FGA in inpatients with PD. Design. This was a prospective cohort study. Methods. One hundred twenty-one inpatients with PD were prospectively enrolled. The FGA and other relevant appraisals of gait, balance, clis' ease severity, and activities of daily living were performed. Six months later, the patients were interviewed by telephone to have their fall information collected. Principal component analysis was used to determine construct validity. Spearman correlation coefficients were used to determine concurrent validity between the FGA and other measures. Cutoff point, sensitivity, specificity, and positive likelihood ratio were calculated for predictive validity based on the receiver operating characteristic curve. Results. One common factor was extracted for construct validity, which cumulatively explained 64.0% of the total variance. Correlation coefficients for the FGA compared with other measures ranged from .57 to .85. The cutoff point for predicting falls was 18, with sensitivity of 80.6%, specificity of 80.0%, and positive likelihood ratio of 4.03. Limitations. This study was limited by the length of time of follow-up and self-reports of falls without the requirement of a fall diary. Medication adjustment after the FGA evaluation may have led to a different cutoff score for identifying those patients who were at risk of falling. Conclusions. The FGA demonstrated good construct validity in patients with PD. It had moderate to strong correlations with other balance and gait appraisals. The FGA can be used to predict falls within the subsequent 6 months.