Problem solving is lauded as beneficial, but students do not all learn well by solving problems. Using the resources framework, Tuminaro J., and Redish E. F., (2007), Elements of a cognitive model of physics problem solving: Epistemic games,Physical Review Special Topics-Physics Education Research,3(2), 020101 suggested that, for physics students, this puzzle may be partially understood by paying attention to underlying epistemological assumptions that constrain the approaches students take to solving problems while working on them. They developed an approach to characterizing epistemic games, which are context-sensitive knowledge elements concerning the nature of knowledge, knowing and learning. As there is evidence that context-activated knowledge influences problem solving by students in chemistry, we explored identifying epistemic games in students’ problem solving in chemistry. We interviewed 52 students spanning six courses from grade 8 through fourth-year university, each solving 4 problems. Using 16 contexts with substance characterization problems, we identified 5 epistemic games with ontological and structural stability that exist in two larger epistemological frames. All of these epistemic games are present at all educational levels, but some appear to grow in across educational levels as others recede. Some games also take lesser and greater precedence depending on the problem and the chemistry course in which students are enrolled and the context of the problem. We analyze these results through a frame of learning progressions, paying attention to students’ ideas and how these ideas are contextualized. Based on this analysis, we propose teaching acts that instructors may use to leverage the natural progressions of how students appear to grow in their capacity to solve problems.