ObjectiveTo verify the association between frailty and motor performance, and identify which test is the best predictor of frailty in the older adults. MethodsCross-sectional, population- and home-based study that analyzed data from 286 older adults. The association between physical performance tests and frailty was verified by binary logistic regression. The diagnostic power of the performance tests and the identification of the best cut-off points to frailty were evaluated using the ROC curve and the area under the ROC curve. ResultsThe prevalence of frailty was 23.8%. Frailty was associated with worse performances in the ‘chair stand’ test (p=<0.01) and the ‘pick up a pen’ test (p=<0.01). Performance requiring more than 14 seconds in the ‘chair stand’ test proved to be a good criterion for discriminating frailty in the older adults, considering the sensitivity and specificity. ConclusionThe worst performances in the tests were associated with the frailty syndrome. All tests were able to predict frailty in the older adults. However, taking into account the high sensitivity of the use of the ‘pick up a pen’ test to confirm a diagnosis of frailty (84.6%), this is recommended for use in home environments.