An image forming apparatus is designed to be capable of easily suppressing a variation of an inclination of a sensor relative to an endless belt and capable of high accuracy and stabilized sensing operation. The image forming apparatus includes a belt unit having an endless belt supported to be rotatable in a circumferential direction, an image forming unit for forming an image on the belt unit, and an optical sensor for detecting light projected onto the endless belt. The image forming apparatus further includes a sensor supporting member for supporting the sensor, and a positioning portion including, as a unit, a first positioning portion for positioning the belt unit by being contacted by the belt unit and a second positioning portion positioning the sensor supporting member by being contacted by the sensor supporting member.