An image-taking apparatus is disclosed which is capable of performing accurate focusing even when a shake occurs. The image-taking apparatus comprises an image-pickup element which photoelectrically converts an optical image formed by an image-taking optical system; a controller which performs a first focusing control based on a phase difference of plural optical images formed of luminous fluxes not passing through the image-taking optical system, which plural optical images are detected by a light receiving sensor that photoelectrically converts the plural optical images, and a second focusing control based on an image signal generated by an output from the image-pickup element; and a shake detector which generates information on a shake. The controller preferentially performs one of the first focusing control and the second focusing control in accordance with the information on the shake.