Tracks scanned by a playback head are provided with a first signal recorded in every nth track, a second signal recorded in a track which precedes and is adjacent to each nth track and a third signal recorded in a track which follows and is adjacent to each nth track. When a preceding track is scanned, the second signal recorded therein is detected; and at a time delayed from the detection of that second signal, the first signal picked up as a crosstalk component is sensed. When a following track is scanned, the third signal recorded therein is detected; and at a time delayed from the detection of the third signal, the first signal picked up as a crosstalk component is sensed. The difference between the first signal sensed when a preceding track is scanned and the first signal sensed when a following track is scanned is indicative of a tracking error.