For two decades, risk management has been gaining ground in banking. In light of the recent financial crisis, several commentators concluded that the continuing expansion of risk measurement is dysfunctional (Power, 2009; Taleb, 2007). This paper asks whether the expansion of measurement-based risk management in banking is as inevitable and as dangerous as Power and others speculate. Based on two detailed case studies and 53 additional interviews with risk-management staff at five other major banks over 2001-2010, this paper shows that relentless risk measurement is contingent on what I call the "calculative culture" (Mikes, 2009a). While the risk functions of some organizations have a culture of quantitative enthusiasm and are dedicated to risk measurement, others, with a culture of quantitative scepticism, take a different path, focusing instead on risk envisionment, aiming to provide top management with alternative future scenarios and with expert opinions on emerging risk issues. In order to explain the dynamics of these alternative plots, I show that risk experts engage in various kinds of boundary-work (Gieryn, 1983, 1999), sometimes to expand and sometimes to limit areas of activity, legitimacy, authority, and responsibility.