Since the publication of the Cadbury Code (1992), there has been a rapid international diffusion of corporate governance codes containing recommendations on boards, executive pay, disclosure and investor relations. But there has been little or no scientific evidence to support the recommendations, so it is questionable whether they can be explained by market failures. Instead, based on public choice theory, I propose the alternative hypothesis that the codes reflect rent seeking by institutional investors in a bargaining game with other stakeholders, including investment banks, auditing firms, incumbent owners, managers and employees. I argue that this hypothesis can explain a number of puzzling features about codes, including their remarkable similarity, the ‘one size fits all’ approach and their pattern of diffusion. A study of fifty-two corporate governance codes and a qualitative review of their core contents provide empirical support for the institutional investor hypothesis. The key focus is on empowering boards and making them accountable to minority investors, i.e., capturing the overall control of listed companies. Since it is not clear that these initiatives contribute to overall value creation, an evaluation of the costs of governance is called for.