A moving-state determining device acquires a first overhead image from a bird's-eye image obtained at first image pickup timing and acquires a second overhead image, which can be regarded as the same road surface area as that of the first overhead image, for each of moving directions assumed for the moving body from a bird's-eye image obtained at second image pickup timing different from the first image pickup timing by a predetermined time period. The device compares a degree of similarity between the first overhead image and each of a plurality of second overhead images to acquire the second overhead image with the largest similarity degree as the same road surface area as that of the first overhead image. The device acquires the moving direction and a lateral displacement amount in the moving from a coordinate conversion condition based on which the second overhead image has been obtained.