This paper raises a proposal for three types of customized training in MT (machine translation) editing to meet the needs of diverse audiences and multiple end uses of MT texts. The proposed MT editing includes discourse-specific, register-specific, and content-specific editing that are tailored to the varied expectations and processing abilities of high-technical, semi-technical and low-technical audiences. Furthermore, to match the variables of trainee translator's translation competence, the audience-oriented proposal outlines a framework of progressive, easy-to-challenging training procedures as functional guidelines for the implementation of effective MT education. In short, different from the ”generalized” editing training, this paper stressed the importance of ”categorized” and ”classified” pre/post-MT editing training to relate academic MT education to the translation needs of the real world.