This paper reports on a work-in-progress that examines a high-stakes teaching and oral language performance assessment known as the TEACH test, in which international teaching assistant (ITA) candidates are required to demonstrate their competency in a panel-rated teaching simulation. This study draws upon the analytic tools of Conversation Analysis to examine 29 videotaped TEACH tests recorded at a large Research 1 (R1) university in the eastern United States with the objective of providing a more in-depth understanding of the demands placed on test-takers and raters. The initial analysis shows how prospective ITAs manage questions that engender problematic responses, and how audience participation can be supportive of the demonstration of competence. In the conclusion, directions for future work are discussed.