Disposing an oscillating plane mirror as a mechanically driven optical system between a first elliptical mirror and a second elliptical mirror would increase overall device size and complicate maintenance. Addressing this the present invention provides a fundus image forming device for scanning the retina of a subject provided with: a first reflective mirror that reflects beam light entering through a first focal point in such a manner that the beam light will pass through a second focal point a two dimensional scanning unit that is disposed at the position of the first focal point of the reflective mirror and reflects entering beam light so as to be swept two dimensionally and a second reflective mirror that reflects beam light entering through a third focal point in such a manner that the beam light will pass through a fourth focal point and is disposed so that the position of the third focal point corresponds to the position of the second focal point of the first reflective mirror. The fundus image forming device is arranged so that the position of the fourth focal point of the second reflective mirror corresponds to the position of the pupil of the subject.