This paper will outline a general methodology for designing evaluations, based on the work of the EAGLES Evaluation Group, which is in turn based on ISO/IEC standards 9126 and 14598. After a brief introduction, a quick recipe for designing an evaluation is given, which is made more concrete first by an informal example from outside the domain of language technology, and then by a more technical example. A more detailed discussion of EAGLES and ISO work is structured around examination of particular technical issues. In particular, it is suggested that three areas urgently require investigation: the solicitation and formulation of user needs, the co-operative definition of quality models for a variety of language technology applications and the definition of commonly agreed valid and reliable metrics for a variety of applications. In conclusion, it is suggested that the methodology outlined could be very widely applied and would constitute a common framework through which those involved in evaluation could communicate, discuss difficulties and compare results.