BACKGROUND AND PURPOSE: The identification of stroke patients who are prone to falls is essential to the development of prevention strategies. This study aimed at developing an easily administered fall risk index for patients in stroke rehabilitation. METHODS: A consecutive series of 135 patients in geriatric stroke rehabilitation was studied. Patient characteristics viewed as potential fall predictor variables were assessed at admission. Univariate and multiple Cox regression analyses of these variables were used in the development of a fall prediction index. RESULTS: The final index included the following items: male sex, poor performance of activities of daily living, urinary incontinence, impaired postural stability, bilateral motor impairment, presence of bilateral cortical or white matter lesions, visuospatial hemineglect, and use of diuretics, antidepressants, or sedatives. The index score correlated significantly with the fall risk (odds ratio, 1.46; 95% confidence interval, 1.26 to 1.69). The score was also used to classify individuals into low-, intermediate-, and high-risk groups, among which the fall rates differed significantly (log rank statistics, 29.86; P < .001). CONCLUSIONS: An easily administered fall risk index is suggested, which might serve as a basis for prevention strategies among patients in stroke rehabilitation.