Background: Context-based learning approaches are used to enhance students' interest in, and knowledge about, science. According to different empirical studies, students' interest is improved by applying these more non-conventional approaches, while effects on learning outcomes are less coherent. Hence, further insights are needed into the structure of context-based problems in comparison to traditional problems, and into students' problem-solving strategies. Therefore, a suitable framework is necessary, both for the analysis of tasks and strategies. Purpose: The aim of this paper is to explore traditional and context-based tasks as well as students' responses to exemplary tasks to identify a suitable framework for future design and analyses of context-based problems. The paper discusses different established frameworks and applies the Higher-Order Cognitive Skills/Lower-Order Cognitive Skills (HOCS/LOCS) taxonomy and the Model of Hierarchical Complexity in Chemistry (MHC-C) to analyse traditional tasks and students' responses. Sample: Upper secondary students (n = 236) at the Natural Science Programme, i.e. possible future scientists, are investigated to explore learning outcomes when they solve chemistry tasks, both more conventional as well as context-based chemistry problems. Design and methods: A typical chemistry examination test has been analysed, first the test items in themselves (n = 36), and thereafter 236 students' responses to one representative context-based problem. Content analysis using HOCS/ LOCS and MHC-C frameworks has been applied to analyse both quantitative and qualitative data, allowing us to describe different problem-solving strategies. Results: The empirical results show that both frameworks are suitable to identify students' strategies, mainly focusing on recall of memorized facts when solving chemistry test items. Almost all test items were also assessing lower order thinking. The combination of frameworks with the chemistry syllabus has been found successful to analyse both the test items as well as students' responses in a systematic way. The framework can therefore be applied in the design of new tasks, the analysis and assessment of students' responses, and as a tool for teachers to scaffold students in their problem-solving process. Conclusions: This paper gives implications for practice and for future research to both develop new context-based problems in a structured way, as well as providing analytical tools for investigating students' higher order thinking in their responses to these tasks.