Systematic review methods are developing rapidly, and most researchers would recognise their key methodological aspects, such as a closely focussed question, a comprehensive search, and a focus on synthesising ‘stronger’ rather than ‘weaker’ evidence. However, it may be helpful to question some of these underlying principles, because while they work well for simpler review questions, they may result in overly narrow approaches to more complex questions and interventions. This commentary discusses some core principles of systematic reviews, and how they may require further rethinking, particularly as reviewers turn their attention to increasingly complex issues, where a Bayesian perspective on evidence synthesis, which would aim to assemble evidence - of different types, if necessary - in order to inform decisions’, may be more productive than the ‘traditional’ systematic review model. Among areas identified for future research are the examination of publication bias in qualitative research; research on the efficiency and potential biases of comprehensive searches in different disciplines; and the use of Bayesian methods in evidence synthesis. The incorporation of a systems perspective into systematic reviews is also an area which needs rapid development.