Falling has become a key factor that affects the quality of life of the elderly. Currently, the use of a few rehabilitation robots can contribute to the restoration of balance. In this paper, a walker-based rehabilitation robot with a gait-phasedependent control algorithm is proposed to promote dynamic balance in the elderly. It has unique characteristics in that the level of the walker to resist the propulsion force exerted by a user can vary depending on the gait-phase that is estimated using the interaction force between the robot and the user. The robot efficiently improves the muscle power of various muscle groups of the user. Experiments with three young subjects were conducted to validate the effectiveness of the walker with the gait-phase-dependent control algorithm.