An ultrasonic image generating device which generates a cross-sectional image of a specific cross-section of a subject from ultrasonic images obtained by scanning the subject from a plurality of directions using an ultrasonic probe, includes: a cross-section position identifying unit which obtains cross-section information indicating a position and an orientation of the specific cross-section a positional information obtaining unit which obtains positional information including a position and an orientation of each of the ultrasonic images of the subject a reference image selecting unit which selects at least one of the ultrasonic images as a reference image, the at least one of the ultrasonic images having a distance from the specific cross-section that is less than a first threshold and an orientation difference from the specific cross-section that is less than a second threshold and a cross-sectional image generating unit which generates the cross-sectional image using the reference image.