Translation quality assessment (TQA) is an essential part of any theoretical concept of translation. In the present article the author attempts to provide a basis for an informed use of TQA in translator training. His arguments are based on authentic examples of TQA as it is practised and on an overview of frequently used evaluation scenarios. This provides an empirical base for a detailed discussion of theoretical concepts (mainly Gutt's relevance concept and Kussmaul's functional approach). The first part illustrates that popular concepts of translation are not built on a tabula rasa. Laypersons hold dogmatic positions, particularly those of 'faithfulness' and 'cultural neutrality'. These are shown to be basically functionalist concepts, albeit not recognised as such. Gutt's criticism of translation theory in general and functionalism in particular is discussed. The second part provides an overview of how TQA is carried out (in Germany) in various scenarios. This shows that TQA, as practised in the university training of translators, uses the least homogeneous criteria and seems to be neither integrated into theoretical frameworks nor based on requirements typically made by clients. The author argues that the only way out of this pedadgogical dilemma is to use TQA as a means and not as and end.