In this paper the authors present a statistical test for testing the strict avalanche criterion (SAC), a property that cryptographic primitives such as block ciphers and hash functions must have. Random permutations should also behave as good random number generators when, given any initial input, its output is considered part of a pseudorandom stream and then used as an input block to produce more output bits. Using these two ideal properties, we construct a test framework for cyptographic primitives that is shown at work on the block cipher TEA. In this way, we are able to distinguish reduced round versions of it from a random permutation.