A research experiment was conducted to investigate how accurate a voice recognition system must be for daily production use. Specifically, the purpose of the research was to establish the percentage accuracy level at which a user becomes frustrated and decides not to use a voice recognition device. The experiment consisted of controlling the perceived recognition accuracy of a voice recognition system and then collecting data through the use of a questionnaire from the experimental users on the acceptability of the equipment. The experiment was not totally successful for a variety of reasons. This paper will discuss the research methodology, review the data collected, and suggest possible alternatives to the experimental design to overcome the problem areas encountered. (Author