The invention relates to a method of detecting duplicates in a synchronization system comprising at least one client device and at least two synchronization servers. The method comprises the steps of allocating first identifiers to the data units to be synchronized for use in the client device and second identifiers for use in the synchronization server; maintaining mapping tables for the synchronization servers by adding to the mapping table first identifiers and second identifiers allocated to the data units to be synchronized via the synchronization server. The method further comprises the steps of comparing the identifier pairs formed by the first and second identifiers added to the mapping tables of different synchronization servers; and causing at least part of the identifier pairs of the mapping tables to correspond to each other.