This paper focuses on diagnosis of dynamic faults in SRAMs. The current techniques for fault diagnosis are mainly based on the signature method. Here, we introduce an extension of the signature scheme by taking in account additional information related to the addressing order during March test execution. A first advantage of the proposed approach is its capability to distinguish between static and dynamic faults. Another main feature is the correct identification of the location of the failure in a given block of the memory: the core-cell array, write drivers, sense amplifiers, address decoders and pre-charge circuits. Moreover, since this approach does not modify the March test, there is no increase of test complexity, conversely to other existing diagnosis techniques.