Criteria which are currently used in testing understanding and understandability are critically discussed. It is shown that, while some of them are inadequate, most of them can be used, if the user is able to analyze the kind of result they actually deliver. The analysis should be done in terms of a theory of understanding.