Project management is known for its tools and techniques that are used to plan and deliver projects in a controlled context. Megaprojects don't always fit well into this paradigm due to their size, complexity and longevity. Megaprojects often start without precisely defined goals and without a detailed knowledge of how the project will progress or the outcomes will be delivered. We examine the requirements for governance of megaprojects by reviewing the literature and reflecting on practice. We use the analytical model of where, how and what to illustrate different units of analysis (i.e., context, governance and goals) in megaprojects in three countries and to illustrate how goals and governance move. Building upon the governance and performance management literature, the paper contributes to the understanding of moving goals and governance for ensuring performance. We propose a framework for diagnosing goals and we list six systemic errors that result in a misfit.