In an image capturing apparatus having an image sensor including a plurality of unit pixels each having a plurality of photoelectric conversion portions and a microlens, the image sensor can be scanned with a first scan method of adding and reading out signals from a portion of the photoelectric conversion portions by a predetermined number of unit pixels and with a second scan method of adding and reading out signals from the photoelectric conversion portions by the predetermined number of unit pixels. A pixel signal read out with the second scan method is selected if a defocus amount is larger than a threshold value and the larger of the pixel signal read out with the second scan method and a signal obtained by using the pixel signal read out with the first scan method is selected if the defocus amount is equal to the threshold value or less.