An image pickup apparatus which is capable of properly detecting and compensating for flash bands to generate a corrected image with no different levels of luminance. Based on an image signal output from an image pickup device which sequentially starts exposure and reads out signals for each row of pixels, a flash band appearing in a plurality of frames consecutive in terms of time due to an external flash is detected. Gains are changed with rows of the frames. At least one of the frames in which the flash band was detected is corrected to a full-screen flash image. When flash bands caused by a plurality of external flashes fired in one frame is detected, a gain is calculated based on a light quantity ratio between the external flashes, and a region where the flash bands overlay each other is multiplied by the gain to generate the full-screen flash image.