We report on a design-based research study that was conducted over three iterations. It chronicles the design, development, and implementation of School Scene Investigators, a forensic science game series for middle school students that utilizes mobile augmented reality. Played on mobile devices while exploring the school environment, School Scene Investigators embeds scientific practices in a real-world context. Students work collaboratively playing unique, interdependent roles as they collect and analyze scientific data in order to solve a mystery. School Scene Investigators aims to (1) engage students through the experience of flow, a positive psychological state often experienced during well-designed games and (2) trigger science interest. In order to better understand how to design mobile game environments that engage students in flow and trigger their interest in science, we analyzed students' self-reports of flow and interest after playing the game. Previous research demonstrated that each iteration of School Scene Investigators engaged students in a substantive flow-like experience. In this study, since engagement does not guarantee interest, we tested whether such engagement, measured as flow, was predictably related to triggered science interest. Data were pooled from all three iterations into a Bayesian multilevel model. Findings demonstrated that students with higher flow had a higher probability of triggered interest. Implications for the findings are discussed.