An image capturing apparatus comprises an optical system, an image sensor having pixels each including a plurality of photoelectric converters capable of outputting image signals independently, a driving unit which controls driving of the image sensor, a focus detection unit, and an addition unit which adds the output image signals on a per-pixel basis. In each pixel, the photoelectric converters are divided into groups each including at least two photoelectric converters and a charge accumulation period for one group is delayed from and partially overlaps a charge accumulation period for another. The driving unit drives the image sensor so that image signals are read from the groups in turn, and the focus detection unit detects a focus state using a phase difference method based on the read image signals independently output from the photoelectric converters.