An image-taking apparatus is disclosed which allows a selection of a type of image stabilization to be used from optical image stabilization and electronic image stabilization as required to provide an excellent output image. The image-taking apparatus has an image generator which is operable in a first mode in which it performs electronic image stabilization processing based on a plurality of images sequentially taken with an image-pickup device to generate an output image and a second mode in which it generates an output image without performing the electronic image stabilization processing on an image taken with the image-pickup device. The image-taking apparatus also has a controller which causes image stabilization operation to be performed through driving of an optical system in both of the first and second modes and selects one of the first mode and the second mode.