Abstract: A new method for Software Fault Tolerance (SFT) is presented, which is based on the Data Flow Analysis (DFA).The normal SFT technology is introduced at first, and data errors are pointed out to be the ultimate reason and also the final result of the software faults in the control systems, soit is crucial to adopt a powerful measure for data fault tolerance.Then, the applications of the DFA technique in the SFT are discussed in details, and two important theorems are also given, one for the minimal sufficient rollback point andanother for the checkpoint data range.Besides, the sufficient condition for thevalidation of the definition of a Rollback Block is proposed.The usage of the presented method is illustrated with the telecommunication system as an example.The presented method can be used widely by extending it simply.

Key words: fault-tolerance techniques, finite state automate, data flow analysis, telecommunications software