In a hard disk drive, a two-level pipelined method of processing data errors which minimizes disk re-reads. In the first level, a sector is read and a syndrome is generated based on the ECC bits appended for that sector. If there are no errors, the next sector is read. However, if an error is detected, a flag is checked. If the flag is currently set, the disk drive is prevented from reading the next sector. If the flag is not set, the disk drive is allowed to proceed with reading the next sector; the syndrome is stored in a register; and hardware error correction is performed. If the error is corrected by the hardware circuit, the process repeats by reading the next sector. But if the hardware circuit cannot correct the error, a second level firmware error correction is eventually initiated for handling the error. The flag is set in order to prevent a subsequent syndrome from overwriting the syndrome currently in the register. The flag is cleared only after the syndrome currently in the register has been read by the firmware. Once the flag is cleared, the disk drive may resume reading the next sector, etc.