OBJECTIVE: To investigate the predictive validity of simple gait-related dual-task (DT) tests in predicting falls in community-dwelling older adults. DESIGN: A validation cohort study with 6 months' follow-up. SETTING: General community. PARTICIPANTS: Independently ambulant community-dwelling adults (N=66) aged ≥65 years, with normal cognitive function. Sixty-two completed the follow-up. No participants required frames for walking. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Occurrence of falls in the follow-up period and performance on primary and secondary tasks of 8 DT tests and 1 triple-task (TT) test. RESULTS: A random forest classification analysis identified the top 5 predictors of a fall as (1) absolute difference in time between the Timed Up & Go (TUG) as a single task (ST) and while carrying a cup; (2) time required to complete the walking task in the TT test; (3 and 4) walking and avoiding a moving obstacle as an ST and while carrying a cup; and (5) performing the TUG while carrying a cup. Separate bivariate logistic regression analyses showed that performance on these tasks was significantly associated with falling (P<.01). Despite the random forest analysis being a more robust approach than multivariate logistic regression, it was not clinically useful for predicting falls. CONCLUSIONS: This study identified the most important outcome measures in predicting falls using simple DT tests. The results showed that measures of change in performance were not useful in a multivariate model when compared with an "allocated all to falls" rule.