When a first scheme (transition observation time optimization scheme) is selected, a computation unit computes an actual transition observation time as a time of an integer multiple of a storage-ejection time of a collision cell within a frame of a transition observation time. When a second scheme (storage-ejection time optimization scheme) is selected, the computation unit divides the transition observation time by a maximum storage-ejection time to determine a number of repetitions of a storing-ejecting operation of the collision cell, and determines a storage-ejection time based thereon.