A tandem image forming device for an image forming apparatus of the present invention includes a plurality of image forming means arranged side by side along an intermediate image transfer belt. The image forming means each include a developing device and a cleaning device arranged around an image carrier. One of nearby image forming means has its cleaning device positioned above the developing device of the other image forming means, e.g., an agitating section forming part of the image forming means. This configuration insures smooth circulation of a developer and therefore obviates an irregular developer distribution while reducing the overall size of the image forming device.