Evaluation has always been central to the consciousness of those involved in the field of Machine Translation (MT). Historically, evaluation has proved difficult, traumatic, at times misleading, but very often both revelatory and helpful. Originally, it was the apparent results of evaluations themselves which made the general public aware of the potential for MT. Today, there is emerging a legacy of actual production use of the output of MT from which the fuller understanding of its potential becomes apparent to actual users. The increased usage of MT, however, demands more comparability and relevance among the many attributes and measures of MT. This chapter explores evaluation, to come away with an idea of why it is so central to MT, why it is so difficult to do, and why there must be many different types of evaluation for many types of users and uses.