Concentrated distribution of international reserves is puzzling. I show that the growth rates of international reserves bear only a very weak relationship to their initial stocks (scaled by GDP or in absolute terms), and that, by implication, the cross-sectional distribution of reserves conforms to Zipf's law. The law states that the size of reserves is inversely related to their ranking. Evidence in favor of the law is strong and time robust. I compare the crosssection distribution of international reserves embedded in the WEO projections to that implied by Zipf's law and find that international reserves are much less concentrated in the WEO projections than implied by Zipf's law.