This year marks the 200th anniversary of the publication of Thomas Malthus’s famous essay postulating that human population growth would outstrip the earth’s food-producing capabilities. His writing sparked a debate that has waxed and waned over the last two centuries but has never disappeared completely. Stated simply, Malthus’s proposition was that because population grows exponentially while food supplies expand linearly, the former would eventually outpace the latter. He predicted that hunger, disease, and famine would result, leading to higher death rates. One of the missing pieces in Malthus’s analysis was the power of science and technology to boost land productivity and thereby push back the limits imposed by a finite amount of cropland. It was only in the twentieth century that scientific research led to marked increases in agricultural productivity. Major advances, such as the large-scale production of nitrogen fertilizers and the breeding of high-yielding wheat and rice varieties, have boosted crop yields and enabled food production to rise along with the world population (Dyson 1996). Between 1950 and 1995, human numbers increased by 122% (US Bureau of the Census 1996), while the area planted in grain expanded by only 17% (USDA 1996, 1997c). It was a 141% increase in grainland productivity, supplemented with greater fish harvests and larger livestock herds, that allowed food supplies to keep pace with population and diets for a significant portion of humanity to improve. Despite this remarkable success, concern about future food prospects has risen in recent years because of a marked slowdown in the growth of world grain yields, combined with an anticipated doubling of global food demand between 1995 and 2025 (McCalla 1994, FAO 1996). Whereas annual grain yields (expressed as threeyear averages) rose 2–2.5 % per year during every decade since 1950, they registered growth of only 0.7% per year during the first half of the 1990s (Brown 1997, USDA 1997a, 1997b). Excluding the former Soviet Union, where the political breakup and economic reforms led to large drops in productivity, global grain yields increased an average of 1.1% per year from 1990 to 1995, approximately one-half the rate of the previous four decades (Brown 1997). Today, the principal difference between those analysts projecting adequate food supplies in 2025 and those anticipating significant shortfalls is the assumed level of productivity growth—specifically, whether annual productivity over the next three decades is likely to grow at closer to the 1% rate of the 1990s or the 2–2.5% rate of the previous four decades. Water—along with climate, soil fertility, the choice of crops grown, and the genetic potential of those crops— is a key determinant of land productivity. Adequate moisture in the root zone of crops is essential to achieving both maximum yield and production stability from season to season. A growing body of evidence suggests that lack of water is already constraining agricultural output in many parts of the world (Postel 1996, UNCSD 1997). Yet to date, I am aware of no global food assessment that systematically addresses how much water will be required to produce the food supplies of 2025 and whether that water will be available where and when it is needed. As a result, the nature and severity of water constraints remain ill defined, which, in turn, is hampering the development of appropriate water and agricultural strategies. In this article, I estimate the volume of water currently consumed in producing the world’s food, how much additional water it will take to satisfy new food demands in 2025, and how much of this water will likely need to come from irrigation. I then place this expected irrigation demand in the context of global and regional water availability and trends. Finally, I discuss the policy and investment implications that emerge from the analysis.