This chapter discusses some problems from which pronunciation training classes at universities often suffer. As a starting point it is argued that such classes should be laid out, to a greater or lesser extent, based on articulatory phonetics and that a practice-makes-perfect approach does not work, at least for adult learners in that it fails to facilitate matured cognitive abilities. However, given the extremely limited time available under the semester or quarter system, the instructor has to decide not what to include but rather what to exclude. It is proposed in the development of this discussion that the course should be structured so that the acquisition of the English prosodic system is prioritized, with the proper knowledge of nuclear stress placement being the ultimate goal. The treatment of segments should be restricted only to those segments which Japanese learners find hard to produce or perceive. Finally, a realistic model plan is introduced together with prosodic and segmental items to be included.