The study presented in this article examined the impact of two simulation-based inference activities on students' understanding of p-values in a second undergraduate statistics course. In the study, students familiar with traditional inference methods used physical and computer simulations to estimate p-values. To examine students' conceptions before and after participation, students were prompted to write brief explanations of p-values, to apply these explanations in different contexts, and to self-evaluate their understanding. This article describes students' varying conceptions of p-values and the ways in which students demonstrated improvement after engaging in simulation-based inference activities. Qualitative analysis of student responses suggests that students hold a range of conceptions of p-values even after a full semester of instruction; further, this study suggests that incorporating simulation-based inference into traditional courses has the potential to improve understanding of p-values and is a promising area for future research.