This paper addresses the general question of the role of emotions in risk management, with an application to the siting of a repository for spent nuclear fuel. Although it is true that such siting has been very difficult to carry out without meeting with strong local opposition, there are some exceptions in recent Swedish experience. This recent experience constitutes an occasion for reconsidering the received message from risk perception research since the 1970s. This research has purportedly demonstrated a very strong impact of emotional processes on risk attitudes, which in turn is supposed to imply that these attitudes are rigid and unchangeable. However, when this assertion is looked at more closely, it is found that a number of questionable assumptions and overly strong generalizations have been made. The central role played by the concept of ‘dread’ is particularly important. Dread, or a ‘gut reaction’, is by no means demonstrated in the traditional selection of hazard attributes, which contain such dimensions as concern for future generations or voluntariness. Such dimensions may have a relationship to emotional processes but they are not emotional per se and their emotional significance remains to be documented. The items that do measure emotional reactions directly, by asking about fear and anxiety, are found in data on nuclear waste repository siting to have only a weak influence on policy attitudes. The most important weight is instead carried by items measuring the ‘expected severity of consequences’. The argument is, in a concluding section, carried to the level of attitudes to technologies, which may or may not be related to perceived risk. It is found that an important role is played by substitutability of a technology. If people see no good substitute for a technology that has important benefits, they are ready to accept the risks. The current situation in Sweden with regard to nuclear power can be understood in this light, since no realistic alternatives are available.