Natural disasters, such as the 2004 East Asian Tsunami, attract a high level of donations. Previous literature has shown that the scale of the disaster is important in driving the aid response, but there are inconsistent findings on whether the number killed or the number affected matters more. In this paper we discuss a number of issues in linking measures of the scale of a disaster to the aid response, particularly taking account of outliers in both scale of disaster and aid. We show that a log-specification is preferred and that this specification can reconcile findings based on different datasets. Both the number killed and the number affected matter equally for whether aid is given; the number killed is more strongly related to the magnitude of the aid response. We also present new evidence confirming the importance of publicity for disasters, focusing on appeals.