From the experiential learning perspective, this study investigates middle and high school students (n = 1009) who used an online module to learn about wildfire hazards, risks, and impacts through computational simulations of wildfire phenomena. These students were taught by 18 teachers in urban, rural, and suburban schools across the United States. We analyzed students' simulation behaviors captured in log files, responses to an assessment administered before and after the module, and demographic surveys, as well as teachers' responses to a post-module implementation survey. Using mixed effects generalized linear modeling, we investigated whether students' simulation experiences, their prior real-world wildfire experience, and the strategies used by their teachers predicted their understanding of wildfire concepts. In estimating the effect sizes of these variables, we controlled for student variables such as gender, race, English language status, prior wildfire knowledge (pre-test), and module completion rate. Results indicate that students' simulation experience and teacher variation were the two most significant effects, followed by students' real-life wildfire experience. Teacher variations were further explained by differences in teachers' pedagogical strategies while implementing the module. Implications of these findings are discussed for the design and further research of simulations used as proxies for experiential learning of natural hazards.