The present invention enables a subject to hold breath at an arbitrarily defined position and can confirm whether the breath is really held while an image is being captured and reflect the result thereof in an image capturing sequence. An image (411) that displays the respiratory displacement over time of a subject is generated from a nuclear magnetic resonance signal acquired by repeatedly performing a navigation sequence (231) on the subject and displayed. A confirmation operation when an operator who has been seeing the image has confirmed a breath-hold state is accepted from the operator, switching from the navigation sequence (231) to a real image capturing sequence (211) is performed, and the real image capturing sequence is executed for a predetermined period. This process is repeated multiple times. Moreover, it is possible to determine, from the difference in respiratory displacement between immediately before and immediately after the real image capturing sequence, whether breath hold during the real image capturing sequence has succeeded or not.