There is a growing interest in the role played by attention and cognitive processing in the control of posture and gait, investigated by means of dual task methodology. The rationale for dual task testing is to assess whether the simultaneous performance of two (or more) attention- demanding tasks exceed the central information-processing capacity. Recently, we proposed a new ambulatory system for gait analysis (Physilog®) and validated its performances in studying normal and abnormal walking [1]. This system measures temporal and spatial parameters of gait during long period of walking and therefore provides stride-to- stride variability of these parameters. In this pilot study, we used the Physilog® system to assess, under both single and dual task conditions, gait changes in 6 elderly subjects after a ten- weeks balance training program. Our objective was to determine whether dual task testing would be more sensitive than single task testing in detecting gait improvement in these elderly persons