The use of causal diagrams to externalize the mental representation of a problem is recognized to be an important step in solving complex problems. In geography education several global challenges taught about in class are highly complex due to the interconnectedness of many causes and consequences. A systems thinking approach might be helpful to better understand these global challenges. Former studies have shown the effectiveness of concept maps and causal diagrams to foster students' systems thinking. However, it is not always obvious for students to construct proper causal diagrams. In order to optimize teaching strategies concerning these complex systems in geography education, this study analyzes students' cognitive strategies while constructing a causal diagram. We used task-based thinkaloud interviews to study their cognitive strategies. Four different cognitive strategies were observed. The different types of cognitive strategies all resulted in an acceptable constructed causal diagram by the students. The presented insights are explorative, but it reveals the thinking processes that are mostly tacit and therefore has the potential to contribute to better teaching strategies. After all, if we know what processes novices go through while carrying out a complex skill, which are often taken for granted by experts, in this case geography teachers, we can raise awareness among teachers to Theoretical Background Research on Complex Problem Solving Complex global challenges or problems such as climate change, hunger and migration flows are studied in geography. These problems can be considered to be complex ill-structured problems. Jonassen (2004) identifies four ways in which problems can vary: structuredness, domain specificity, complexity, and dynamicity. Solutions to ill-structured problems are not predictable, the procedure to solve the problem is often not known and concepts and principles from different domains are required to solve the problems. Furthermore, these problems are also highly complex and dynamic, due to the amount of involved interconnected variables and the fact that these variables, and thereby the problem as a whole, change over time (Jonassen, 2004). According to these criteria the problems mentioned above are ill-structured problems, also known as wicked problems. Complex problems require different intellectual skills and cognitive operations to deal with than well-structured problems (Funke, 2010; Shin, Jonassen, & McGee, 2003), and therefore teaching strategies should be different (Jonassen, 2004). Several authors describe cognitive strategies involved in complex problem solving. Weinstein and Mayer (1986), for instance, mentioned selecting important information in texts, relating content to already acquired knowledge, structuring the learning content in summaries and schemas, critically processing the content by forming an opinion about explicitly take those processes into account while designing lessons.