A three-dimensional image-capturing device includes a first camera, a second camera, a driving member and a sensing member. The second camera is at one side of the first camera and is capable of being moved laterally by the driving member. After depth-sensing and establishing orientations in an image, the driving member is able to drive the second camera to move relative to the first camera, to change a distance between the first camera and the second camera. Such relative movement allows a greater range of three dimensionality for face recognition and other purposes.