A method for restoring data is provided in embodiments of the present invention, wherein currently processed data set is divided into n data blocks and m redundant coding blocks of the data set are calculated. The method also comprises: analyzing locations, of broken data of broken blocks in the n data blocks, in each broken block when there are broken blocks in the n data blocks; on the basis of the analyzed location, dividing, according to the same dividing rule, each block of the n data blocks and m coding blocks into h small blocks; among all the small blocks of the n data blocks and m coding blocks, putting small blocks which are in the same location in each data block and coding block together as one subset; for a subset with broken small blocks, using data in unbroken small blocks of the subset with the broken small blocks to restore broken data in the broken small blocks of the subset individually. Accordingly, an apparatus for restoring data is also provided in embodiments of the present invention. With embodiments of the present invention, reliability of data storage can be improved.