This study was devised to check the feasibility and validity of a rating scale specifically designed to evaluate gait impairment in Parkinson's disease (RSGE). Demographic data, a brief questionnaire on general aspects influencing gait and mobility, a battery of scales (Barthel Index; Hoehn and Yahr staging; and Northwestern University Disability, Schwab and England, and Unified Parkinson's Disease Rating Scale [UPDRS]), and timed tests ("Up and Go" and "Steps x Seconds" tests) were recorded under protocol, as was the RSGE-Version 1.0 (23 items in four subscales). Fifty patients enrolled at two centers were included. Twenty-five (50%) were simultaneously (though independently) evaluated by three examiners, in order to determine the interrater reliability. The mean age of the patients was 67.6 +/- 11.16 years, with a mean 8.18 +/- 5.58 years of disease duration. Motor fluctuations were present in 48% of patients. The RSGE Cronbach's alpha was 0.94. Only the item "Dyskinesias" was not correlated with the RSGE total sum. The item "Axial rigidity" showed a fair interrater reliability (kappa = 0.30). However, most of the RSGE items (16/23, 70%) had kappa > or = 0.65. The convergent validity with the applied scales was very high (Spearman r = 0.74-0.90, p < 0.001). The highest correlation (0.90) was obtained with the UPDRS. Also, the RSGE correlation with timed tests was very satisfactory ("Up and Go" = 0.81; "Steps x Seconds" = 0.70; both, p < 0.001). Factor analysis of the RSGE disclosed four dimensions explaining 68% of the variance. The RSGE-Version 1.0 proved to be a valid instrument. The reliability of some items has to be improved, however.