A three-dimensional image-capturing apparatus includes an image-capturing device having a plurality of image-capturing regions and a plurality of optical systems for forming images of a subject in the image-capturing regions. The optical systems includes a plurality of reflectors for reflecting rays from the subject a number of times and at least a lens provided to be closer to the image-capturing device than the reflection means closest to the subject. The reflectors and the lens are used to form, in the image-capturing regions, separate images of the subject which are captured from different viewpoints.