An image processing apparatus that is capable of determining a document size with high accuracy, even when a device for detecting the document size is not placed therein and an input operation of the document size by a user is eliminated is provided. A page background determining section extracts candidates for page backgrounds from input document image data and determines the number of page background types. A white pixel counting section counts the number of white pixels from input document image data. A document size determining section determines the document size of the input document image data, based on the number of page background types determined by the page background determining section, a page background density value which is the lowest density value in one or a plurality of page background areas determined to be page backgrounds, and the number of white pixels extracted by the white pixel counting section.