Risk is now being widely used both as a way of understanding policy and decision-making, and as a way of making decisions. However, there is little agreement on how risk is defined. For some risk is objective and measurable, while for others risk is subjective. What this article demonstrates is that because risk is a contested concept, it is extremely difficult to use it either as a tool for analysing government or for making decisions. In their different ways both scientists and social theorists assume an objectivity to risk. However, risk is not objective but contingent, and depends on decisions that are often related to issues of power. Consequently both governments and analysts are caught between matching subjective and political notions of risk with objective risk assessment. Two case studies are used, BSE in cows and British exchange rate policy, to demonstrate the difficulties in using risk as a way of analysing the policy process and for making decisions.