Walking is an automated rythmic motor behavior that is mostly controlled by subcortical brain regions. Automaticity implies that gait can be performed without attention. However, recent works highlight the involvement of attentional resources in gait, using a "dual-task" methodology in which performance on attention-demanding tasks such as spoken verbal response and walking is compared when they are performed separately and concurrently. Changes in gait patterns due to simultaneous performance of an attention-demanding task are interpreted as interference caused by competing demands for attentional resources involving the cortical level in gait control. Dual-task related gait changes are a new way to assess age-associated change in gait control. Furthermore, new screening tools of falling risk based on the dual-task paradigm have been developed, comparing walking performance alone to walking while performing a simultaneous attention-demanding task. However, the consistent prediction of falls by dual-task testing remains difficult. The recent availability and growing number of validated, user-friendly portable gait analysis systems allow simple objective gait measurement such as gait variability. The study of gait variability under dual-task represents a new challenge for the clinicians because high stride-to-stride variability is a powerful fall predictor in older adults. Because there is increasing evidence that age-related gait changes are associated with incidence of dementia, dual-task related gait changes could provide useful information about relationship between gait disorders and cognitive decline. Furthermore, dual-task could be a new approach of the gait disorders rehabilitation.