Randomised controlled trials (RCTs) have been promoted as a means of improving policy-making by testing new policies. While testing before full-scale roll-out is commendable, this article discusses the challenges of using RCTs in contemporary health policy-making in England. There are at least two challenges in particular that are currently underrepresented in the debate. The first arises from the complexity of many policies which are often too diffuse and unclear in focus to allow for the clear distinction between a policy ‘mechanism’ and its context to be drawn that is required for a RCT. The second challenge relates to the timing of RCTs, which tend to take place either too early in the life of a policy to be meaningful, or too late to have an effect on policy formulation. The authors therefore encourage policy-makers and researchers to be clear about the types of uncertainties ‘field experiments’ are meant to address, which may be addressed better by other types of knowledge generation.