AbstractIn this study, we evaluated the validity of the Braden scale in assessing the risk of pressure ulcers. Longitudinal clinical data including weekly Braden scale scores for 1,138 patients admitted to a university hospital who developed pressure ulcers during the hospital stay and 4,794 who did not develop pressure ulcers were extracted from the hospital's electronic medical record system. Braden scale scores at three points during hospitalization were analyzed: the initial score at admission, the last score recorded before diagnosis (for pressure ulcer patients) or before discharge (for those without pressure ulcers), and the minimum (highest‐risk) score recorded. Using these data, the predictive validity of the scale was evaluated using a cut‐off score of 18, followed by an evaluation of the relative advantages and disadvantages of cut‐off scores from 12 to 19. Among patients in the general units, the minimum score had the greatest sensitivity (0.85), negative predictive value (NPV; 0.98), and Youden index (0.73). Among patients in the intensive care units, the last score had the best NPV (0.65), Youden index (0.53), and area under the receiver operating characteristic curve (0.78), while the minimum score had the highest sensitivity (0.88). The optimal cut‐off score for patients in the general units was 19 and for those in the intensive care units was 18. These results support a higher cut‐off score than previously recommended, particularly for severely ill patients who are more prone to developing pressure ulcers. © 2014 Wiley Periodicals, Inc.