The relationship between comminution testwork results and the milling circuit that is subsequently selected is not a direct one and frequently involves recourse to databases of past test results. This in itself reflects the lack of a systematic engineering approach in the design process. This paper reviews comminution tests for SAG mill design in an attempt to identify where the essential information comes from that will be used in mill selection. The conclusion is that results from traditional test methods, including obscure and little used procedures like the Mokken test, are more valuable from an engineering perspective than parameters for breakage rate based modelling. Suggestions are made on ways in which existing test procedures can be enhanced in order to generate parameters that can be directly related to characteristics of the milling circuit being designed.