Graduate teaching assistants (GTAs) are responsible for direct instruction of geoscience undergraduate students at an array of universities and have a major effect on the knowledge, beliefs, and practices of their students. GTAs benefit from in-department training in both beliefs and practices that align with the existing literature on teaching and learning in the discipline, and such training can have long-standing effects when GTAs transition into faculty roles. However, the most recent review, in 2003, revealed little literature examining outcomes of geoscience GTA training programs. Using the framework of the GER Strength of Evidence Pyramid, this article outlines the development and application of a rubric to allow the user to analyze the existing geoscience GTA training literature and provide example study designs at each level of strength. Extending back to 1980, we discovered a total of three peer-reviewed articles describing and empirically evaluating the effect of GTA training programs in the geosciences. Thus, this article also draws from other science disciplines to provide examples for the levels of the rubric not currently represented in the geoscience literature, providing a set of contextually similar models that future designers of geoscience GTA training might draw on to maximize their strength of evidence, given specific institutional and programmatic constraints. Furthermore, we describe ways in which the use of the rubric provides a framework for characterizing the GTA training literature, which revealed areas of research and characteristics of rigor needed for future work.