High stakes examinations can have profound implications for how science is taught and learned. Limitations of school science such as the 'cookbook problem' can potentially be addressed if high stakes assessments target learning outcomes that are innovative. For example, less mindless procedural engagement and more thoughtful consideration of practical science can potentially improve science learning. In this paper, we investigate how practical work is represented in the assessment frameworks of several countries that demonstrate above average performance in the latest PISA science assessments. The main motivation is the need to understand if there are aspects of high stakes summative assessments in these countries that can provide insight into how best to structure national examinations. Assessment documents from a set of selected countries have been analysed qualitatively guided by questions such as 'what is the construct of practical science' and 'what is the format of assessment?' The examined jurisdictions used different approaches from traditional external pen-and-paper tests to internal teacher assessments that included different formats (e.g. laboratory report). Innovative approaches to the assessment of practical skills (e.g. PISA computer-based tasks) do not seem to be represented in these high-stakes assessments. Implications for innovative assessments for high-stakes purposes are discussed.