How can we understand the results of a simulation study? In this article, I address this epistemic question for social science simulations. I argue that we can distinguish two categories of simulations: simulations S_TE, which possess key features that resemble the epistemology and methodology of thought experiments, and simulations S_E, which resemble the epistemology and methodology of experiments. Based on Woodward's theory of causal explanation, I put forward the hypothesis that S_TE provide more understanding and a different kind of knowledge than S_E because they give well-founded answers to what-if-things-had-been-different questions. Epistemic opacity is a persistent problem for simulations S_E, while for S_TE it need not necessarily be so.