The concept of complexity is gaining recognition as a useful framework for theorising and examining multifactorial, widespread health issues. Since it is also an apt descriptive tool for issues that are complicated and multifactorial, exploring the complexity of health systems and public health problems can provide valuable insights about the nature and extent of the challenges faced by policy makers and practitioners. This research set out to explore qualitatively perceptions of responsibility for healthy eating. Data consisted of eight corporate and government documents (on responsibility and health), starting with the 2007 Foresight report on obesity (the first key public health report in the UK to explicitly use a complex systems framework); eight focus groups with a variety of members of the public; and 18 interviews with key representatives at the public health–food system nexus, from the food industry, government, and non-governmental organisations. A Foucauldian discourse analysis was conducted on the dataset to investigate relevant themes. Complexity portrayed diet–health problems, especially obesity, as so multilayered and difficult as to render them insurmountable. Additionally, complexity was routinely used to divert emphasis from diet-specific interventions, such as altering the food environment, to related but conspicuously non-food interventions, such as the sponsoring of physical activity schemes. An unexpected finding was how the notion of complexity in the more usual sense (ie, complicated and multifactorial) emerged from the data as a discursive device. The very act of labelling a system or problem as complex can function discursively to frame it as burdensome and unmanageable. Although this framing may have some validity, it has also become a convenient rhetorical tool—a smokescreen that justifies a lack of policy action to address public health challenges. Within discourse, complexity functions both as a call to unified action, and as a means of justifying inaction by government and the corporate sector. As public health researchers, we must therefore be careful and clear about how we use the term complexity, and the contradiction warrants further investigation. This research was funded by Queen Mary University of London and the London School of Hygiene & Tropical Medicine as a PhD scholarship (to NS).