A test case design technique in which a model is used of the statistical distribution of the input to construct representative test cases.