There are provided a method and apparatus for high-speed compensating repeatable run-out in a hard disk drive. The method includes a) obtaining a value of a position error signal (PES) of the hard disk drive, the PES being a relative position signal of the head with respect to the center of the target track of the disk; b) determining whether the head is in the settling mode; c) if the head is in the settling mode, compensating the RRO error without feeding back the PES; and d) if the head is in the track-following mode, compensate the RRO error by feeding back the PES. Accordingly, an excessive RRO error is compensated. Even though the hard disk drive receives an external shock or is assembled in an offline servo writing manner, the performance of the hard disk drive does not deteriorate.