An image forming apparatus is operable in a first mode where a toner images on one of a plurality of image bearing members are transferred to an intermediate transfer belt; and in a second mode where toner images on the image bearing members are superimposedly transferred to toner images on the intermediate transfer member. A bias controlling unit controls the bias of the intermediate transfer belt by rotating the intermediate transfer belt until the toner images on the image bearing members are transferred to the intermediate transfer belt after the image bearing members have contacted with or separated from the intermediate transfer belt. A duration when the intermediate transfer belt is rotated when the second mode is changed to the first mode is shorter than a duration when the intermediate transfer belt is rotated when the first mode is changed to the second mode.