In the aftermath of the 2006 and 2014 Thai coups, observers declared the resurrection of the bureaucratic polity. Bureaucrats, though, remained influential even during the period of 1992–2006, when elected politicians were thought to command the Thai state. Bureaucratic involvement in politics poses a challenge for dominant political science theories of politician–bureaucrat relationships, which draw heavily from principal–agent frameworks. I apply agency theory to Thailand, testing three different hypotheses derived from the theory. Examining legislative productivity and control over bureaucratic career trajectories, I find that elected politicians increasingly acted as principals of the Thai state from 1992 through 2006, and to a lesser degree from 2008 to 2013. Thai bureaucrats, though, have frequently engaged in the political sphere, blunting political oversight and expanding their independence vis-à-vis politicians. This suggests that the principal–agent model overlooks the range of resources that bureaucracies can bring to bear in developing countries, granting them greater autonomy than anticipated. As such, theories of the politician–bureaucrat relationship in developing states need to better account for the mechanisms through which bureaucrats exercise policy discretion and political influence.