An image processing apparatus capable of efficiently using a removable medium when the removable medium is used instead of a memory provided in the image processing apparatus. The apparatus includes a RAM for storing image data, a removable medium I/F to which the removable medium is coupled, and a CPU. The CPU confirms the remaining capacity of the RAM and determines based on the remaining capacity confirmed during execution of a job and a job operation mode whether the job can be continued by using only the RAM. When it is determined that the job cannot be continued, image data to be saved is selected according to a predetermined priority order, and the selected image data is saved from the RAM onto the removable medium.