An image forming apparatus has a plurality of developing units for developing toner images of different colors, an intermediate transfer belt for transferring the toner images of the respective colors onto a surface of the belt or onto a recording sheet transported over the belt surface, a first or drive roller for driving the belt while the belt is wound around the first roller, a second or driven roller and a tension roller, which is driven along with the driving of the belt driven by the first roller. The image forming apparatus satisfies a relation defined by the following equation: D2=ND1 where D1 is a belt-inclusive outer diameter of the first roller where the belt is wound, including the thickness t of the belt, D2 is an outer diameter of the second roller, and N=n or 1/n where n is a positive integer.