This chapter discusses the evaluation of speech synthesis. It does not attempt to present an overview of all the techniques that may be used, or to cover the full history of previous evaluations, but instead it highlights some of the weaknesses of previous attempts, and points out areas where future development may be needed. It presents the view that speech synthesis should be judged not as a technology, but as a performance, since the actual intended listener presumably has less interest in the achievements of the developers than in the effect of the speech, its pleasantness, and its suitability to the various listening contexts.