An image capturing apparatus includes an image sensor, composition unit, coupling unit, and arithmetic unit. The image sensor has a plurality of pupil-divided focus detection portions each formed from a first focus detection pixel and second focus detection pixel. The composition unit composites signals output from the first focus detection pixels to obtain a first composite signal, and composites signals output from the second focus detection pixels to obtain a second composite signal in each of sections assigned to the image sensor to include a plurality of focus detection portions. The coupling unit couples first composite signals to obtain a first coupled signal, and couples second composite signals to obtain a second coupled signal in the plurality of sections. The arithmetic unit calculates the defocus amount of an image forming optical system on the basis of the first and second coupled signals.