Education in the twenty-first century must prepare students to meet the challenges of a dynamic and interconnected world. However, assessment of students' skills tends to focus primarily on static tasks. Therefore, it is not known whether knowledge about successful strategies displayed on static tasks can be transferred to interactive and dynamic environments. This study investigated whether students' knowledge of a certain strategy (i.e., vary-one-thing-at-a-time, VOTAT) that was assessed in a paper-and-pencil-based scientific reasoning task as well as their fluid intelligence and learning orientation would be sufficient to explain variance in the application of the VOTAT strategy in solving an interactive complex problem solving (CPS) task (i.e., CPS strategy). Furthermore, we analyzed whether CPS strategy mediated the relation between the predictors (i.e., scientific reasoning, learning orientation, fluid intelligence) and CPS performance. The sample consisted of N = 3,191 Finnish students attending the 6th and 9th grades. Results revealed that all predictors were significantly related to CPS strategy, but a substantial amount of variance in CPS strategy remained unexplained (?R[superscript 2] = 0.583). Furthermore, CPS strategy mediated the relation between the predictors and CPS performance. Three implications are discussed: Different demands on the problem solver, knowledge transfer from static versus interactive tasks, or metastrategic knowledge may explain the unexplained variance in CPS strategy. Additionally, the results of our mediation analyses emphasize the importance of measuring strategies via logfiles to gain a deeper understanding of determinants of students' CPS performance. Finally, fostering motivational factors such as students' learning orientation yields small improvements in CPS performance.