A conventional stereo image pickup device forms an appropriate stereo image by changing the stereo base in accordance with the distance to a subject, but fails to select an appropriate focal length or stereo base by simply using the subject distance without obtaining the subject size. A stereo image pickup device (1000) estimates the size of a subject in accordance with the imaging mode set in a typical two-dimensional camera, and estimates the subject distance using the estimated size of the subject and the focal length, and further calculates the stereo base that enables an optimum disparity to be achieved using the subject distance, determines the stereo base, and aligns the two imaging units (the first imaging unit (103) and the second imaging unit (104)). As a result, the stereo image pickup device forms a stereo image having an appropriate stereoscopic effect.