Objective: The first aim of this study was to determine whether ordinary postural clinical performance varied with an additional cognitive task. The second purpose was to valid the predictive value for falling of two clinical tests. Methodology: The sample comprised ninety-five women with osteoporosis who lived independently in the community and aged over 70 years (mean 73.4 ± 1.7 years). Timed “Up & Go” test (TUG) and One leg balance (OLB) were realised. Tests were randomly assigned. Trials were realised with or without a cognitive additional task (math task involving subtraction by twos or fives or addition by threes). Results: For both tests (TUG and OLB), performance decreased in the dual-tasks condition (p<0.05). On the other hand, performances were not different in old women with and without an history of fall whatever the test and with or without an additional cognitive load. Conclusion: Performs in Timed Up & Go test and in One leg of balance were less efficient when simultaneously performing a cognitive task. These tests realised without or with cognitive task did not predict falls in a sample of women who lived independently.