Climate change is likely to impact more severely on the poorer people of the world, because they are more exposed to the weather, because they are closer to the biophysical and experience limits of climate, and because their adaptive capacity is lower. Estimates of aggregated impacts necessarily make assumptions on the relative importance of sectors, countries and periods; we propose to make these assumption explicit. We introduce a Gini coefficient for climate change impacts, which shows the distribution of impacts is very skewed in the near future and will deteriorate for more than a century before becoming more egalitarian. Vulnerability to climate change depends on more than per capita income alone, so that the geographical pattern of vulnerability is complex, and the relationship between vulnerability and development non-linear and non-monotonous. © 2004 Elsevier Ltd. All rights reserved.