Many new technologies are being implemented to enhance training system effectiveness, particularly in phraseology applications such as air traffic control (ATC). One of these technologies, speech recognition, provides both unique challenges to the developer and unique benefits to the user of the system. This paper shows that requirements for good speech recognition and for good overall training are not mutually exclusive. Specific issues will be addressed, including the development of an appropriate training phraseology as well as common concerns of the user. Guidelines for attaining acceptable recognition accuracy will be provided along with some different methods for quantifying accuracy. Examples of air traffic control phraseology are used in this presentation with specific references to the Shore Based Radar ATC Training System (SATS), the Advanced Shipboard ATC Training System (ASATS), and the Tower Operator Training System (TOTS).