Does irrigation really use 70% of the world's fresh water?
“Agriculture uses 70% of the world's fresh water” is one of the most repeated numbers in water policy. It opens reports, speeches and, often, the business case for precision irrigation. How solid is it?
What FAO says today
FAO's State of the World's Land and Water Resources for Food and Agriculture 2025 puts agriculture at 72% of total freshwater withdrawals in 2020, ahead of industry (16%) and services (12%). FAO's 2025 AQUASTAT data snapshot gives a very similar split for 2022: about 71–72% for agriculture, 15% for industry and 13% for municipal uses.
Two details matter. First, these are withdrawals: water taken from rivers, lakes and aquifers. Part of it returns to the environment. Second, the global share is an aggregate. AQUASTAT's own methodology notes explain that the world total (about 69% agricultural at the time of writing) is pulled up by a few countries with very large withdrawals. Averaged country by country, the agricultural share falls to about 59%.
Withdrawal is not consumption
Europe shows why the distinction matters. At the time of the FIGARO project, irrigation accounted for roughly a quarter of water abstraction in Europe, against 44% for power-station cooling. Yet almost all cooling water returns to rivers, while often only about a third of irrigation water does (see our note on agriculture and Europe's water). Measured by water actually consumed, irrigation weighs far more than its abstraction share suggests.
A 2025 audit of the figure
In November 2025, Arnald Puy, Bruce Lankford and colleagues published “Widely cited global irrigation statistics lack empirical support” in PNAS Nexus. They followed the citation trails behind two famous claims: that irrigation provides about 40% of food and uses about 70% of freshwater withdrawals. Their findings:
- only about 1.5% of the documents cited for these figures actually contain supporting data;
- the 70% figure traces back to a 1992 World Resources Institute report built on country-level national statistics;
- the 40% food figure traces back to work by the Soviet hydrologist M. I. L'vovich (1974), which concerned projected grain supply for 2000, not total crop production;
- once the uncertainty is propagated, irrigation's share of withdrawals ranges from 45% to 90%, and its share of crop production from 18% to 50%.
| Source | Agriculture's share | What it measures |
|---|---|---|
| FAO, SOLAW 2025 | 72% (2020) | Global freshwater withdrawals, all agriculture |
| FAO AQUASTAT snapshot 2025 | about 71–72% (2022) | Global freshwater withdrawals |
| AQUASTAT methodology notes | 69% aggregate, 59% country average | Same data, two ways of averaging |
| Puy et al., PNAS Nexus 2025 | 45–90% | Irrigation's share of withdrawals, with uncertainty |
What it means for precision irrigation
None of this says irrigation uses little water: every estimate makes it the largest user. But savings forecasts that multiply a global “70%” by an efficiency gain are less precise than they look. The authors' own recommendation is to favour place-based measurement over global figures. That is the logic precision irrigation already follows: meter what is applied, measure soil moisture and evapotranspiration on the farm, and judge savings field by field, as the FIGARO pilots did (see how the platform worked).
Sources
- FAO (2025), The State of the World's Land and Water Resources for Food and Agriculture 2025, chapter on water for agriculture.
- FAO (12 December 2025), “Renewable water availability per person plunges 7 percent in a decade”, FAO newsroom.
- FAO AQUASTAT, Water use: methodology.
- Puy A. et al. (2025), “Widely cited global irrigation statistics lack empirical support”, PNAS Nexus 4(11), pgaf323, open access.