The invention relates to a file backup method based on fingerprints, belonging to the technical field of storage backup in a computer. The invention is amied at reducing the backup of duplicated data, saving the cost of network bandwidth and storage needed by backup, and improving the backup efficiency. The invention comprises a backup process and a recovery procedure. The invention adopts the file blocking technology based on an anchorage to identify the redundant data of the backup file, the modification is stable, and the computation cost is low; data blocking of the file is memorized in a blocking volume in a storage server by using the fingerprints thereof as an identifier, thereby avoiding the backup of the duplicated data, and facilitating the data blocking to be shared by different files; file metadata and file blocking index are memorized in a storage volume in the storage server, thereby facilitating a user to organize a storage pool, and realizing the personalized management to logical objects.