An image forming apparatus includes a plurality of image carriers, an intermediate transfer belt, a plurality of rollers, a correction mechanism, and a belt cleaning unit. The intermediate transfer belt has an endless shape, and toner images respectively formed on the plurality of image carriers are sequentially overlaid and transferred thereto. The correction mechanism moves one end side in an axial direction of a tension roller along a direction perpendicular to the axial direction, so as to correct meandering of the intermediate transfer belt with respect to the tension roller. The belt cleaning unit includes a cleaning blade that is disposed in such a manner that its tip portion contacts with an outer circumferential surface of the intermediate transfer belt at a position facing the tension roller via the intermediate transfer belt, so as to remove extraneous matter from the outer circumferential surface of the intermediate transfer belt.