Dunleavy's bureau-shaping model has breathed new life into existing debates about the behaviour of senior bureaucrats. This article assesses the utility of this model as an explanation of the development of Next Step agencies in the last decade in Britain, using data drawn from a series of extensive interviews with senior civil servants. Our conclusion is that, although the bureau-shaping model represents a significant advance on previous models of bureaucratic behaviour that stress budget maximization, it is flawed. In particular, we argue that: it pays insufficient attention to the broader political context within which civil servants operate; mis-specifies bureaucrats' preferences; and oversimplifies the distinction between managerial and policy advice work. More specifically, we suggest that any explanation of the development of Next Steps agencies needs to recognize that: politicians rather that civil servants played the major role in their creation; the strategic calculations of bureaucrats were significantly more sophisticated than the model assumes; and the consequence of the reforms has been that senior civil servants have played a greater, rather than a more limited, management role.