[en] In the development of reactor systems, if cost penalties are not too great, direct testing of reactor components is usually undertaken. When the financial burden of such direct testing becomes too severe, ensuring the adequate performance of reactor components entails the use of proven design methods. The establishment of such methods involves: a) the carrying out of basic generic work to determine the physical phenomena and physical principles involved in component performance. b) the use of this generic data in the generation of mathematical models by means of which reactor situations can be assessed. c) the validation of these models against experimental data (produced at a reasonable cost) in fairly complex situations representative of the reactor configurations. This paper describes the present position, in the UK, in respect of the development of such design methods for the CDFR above sodium environment