Introduction Older adults' perception of their own risk of fall has never been included into screening tools. The goal of this study was to evaluate the predictive validity of questions on subjects' self-perception of their own risk of fall. Methods This prospective study was conducted on a probabilistic sample of 772 Spanish community-dwelling older adults, who were followed-up for a one year period. At a baseline visit, subjects were asked about their recent history of falls (question 1: 'Have you fallen in the last 6 months?'), as well as on their perception of their own risk of fall by using two questions (question 2: 'Do you think you may fall in the next few months?' possible answers: yes/no; question 3: 'What is the probability that you fall in the next few months?' possible answers: low/intermediate/high). The follow-up consisted of quarterly telephone calls, where the number of falls occurred in that period was recorded. Results A short questionnaire built with questions 1 and 3 showed 70% sensitivity (95% CI: 56%-84%), 72% specificity (95% CI: 68%-76%) and 0.74 area under the ROC curve (95% CI: 0.66-0.82) for prediction of repeated falls in the subsequent year. Conclusions The estimation of one's own risk of fall has predictive validity for the occurrence of repeated falls in older adults. A short questionnaire including a question on perception of one's own risk of fall and a question on the recent history of falls had good predictive validity.