The paper discusses qualitative input—output methods. It is shown that information is lost. Because the binaiy relationship constructed by qualitative methods is not transitive, the model lacks economic consistency. Qualitative methods are tending to become more sophisticated, but some problems of economic interpretation are raised.