Research investigating graduate teaching assistants' (TAs') knowledge of fundamental statistics concepts is sparse at best; yet at many universities, TAs play a substantial role in the teaching of undergraduate statistics courses. This paper provides a framework for characterizing TAs' content knowledge in a sampling context and endeavors to raise new questions about TAs' content knowledge and its potential impact on the teaching of undergraduate statistics. The participants in this study were sixty-eight TAs from 18 universities across the United States. These TAs demonstrated considerable knowledge of theoretical probability distributions. However, they experienced tensions when attempting to quantify expected statistical variability in an empirical sampling situation and had difficulty explaining conceptual ideas of variability. (Contains 4 figures and 10 tables.)