An embodiment of the present invention discloses a data deduplication method. The method includes the steps of acquiring one current file in data; determining whether a similar historical file exists according to a sampling data block in at least one predetermined position in the current file; responding to historical files that have no similarities, and storing the current file and corresponding metadata in file as a unit; and responding to historical files that have similarities, and deduplicating the current file in block as a unit. According to the method, data deduplication efficiency can be improved.