There is increasing recognition that sustainability science should be solutions orientated and that such solutions will often require transformative change. However, the concrete sustainability interventions are often not clearly communicated, especially when it comes to the transformative change being created. Using food and energy systems as illustrative examples we performed a quantitative systematic review of empirical research addressing sustainability interventions. We use a modified version of Donella Meadows' notion of ‘leverage points’ – places in complex systems where relatively small changes can lead to potentially transformative systemic changes – to classify different interventions according to their potential for system wide change and sustainability transformation. Our results indicate that the type of interventions studied in the literature are partially driven by research methods and problem framings and that ‘deep leverage points’ related to changing the system's rules, values and paradigms are rarely addressed. We propose that for initiating system wide transformative change, deep leverage points – the goals of a system, its intent, and rules – need to be addressed more directly. This, in turn, requires an explicit consideration of how scientific approaches shape and constrain our understanding of where we can intervene in complex systems.