Coffee production and water consumption. Prejudice or real problem?

For decades, this one figure has been circulating: 140 liters of water consumption per cup of coffee. It's copied, quoted, and even written into laws, but rarely questioned critically. The image it creates of coffee is that of a water-polluting machine. Yet the figure is wrong. With some of our producers, the actual consumption is close to zero. Water problems in coffee production do exist – but different ones than those being discussed.

Whether coffee production is environmentally sustainable or not is repeatedly raised in sometimes sensational debates. Too much water consumption, too much monoculture, too much fertilizer use. Some of it is true, but of course there are many counterexamples of how coffee can also be produced differently. As diverse and colorful as coffee roasters make their brands and names, coffee is produced just as diversely. In full sun, in full shade. In monoculture, in polyculture. Wet processed, dry processed, or something in between.

That these are different approaches and require different amounts of water is obvious. Yet the figure of 140 liters of water per cup of coffee persists stubbornly and is often repeated uncritically.

So I want to take a differentiated look at these 140 liters of water consumption per cup of coffee here and understand the calculation behind it. Our coffee partners have shared their numbers with me and to get to the point: the 140 liters are not correct. The sum is as individual as coffee can be.

Where does the number come from?

The often-cited figure of 140 liters of water per cup of coffee is technically correct but deeply misleading. Around 96% of these 140 liters are rainwater that falls on the coffee plant and evaporates. This is water that would have fallen on the same soil even without coffee cultivation and would have evaporated from natural vegetation. According to Revolve, only about 1.4 liters per cup would actually come from irrigation (groundwater or surface water). About four liters would correspond to the theoretical dilution requirement for fertilizer residues.

I'm always amazed how uncritically these 140 liters are copied without being questioned. I see nothing more than sensationalism in this. In that sense, it's not surprising to find that the figure comes from a 2003 calculation based on climate data from capital city weather stations, makes no regional differentiation, and equates rainwater with irrigation water.

In 2003, Dutch scientists Hoekstra and Chapagain calculated the 140 liters of water per coffee cup in their frequently cited report «the water needed to have the Dutch drink coffee». In their calculations, they use climate data from the FAO (CLIMWAT). However, this has a catch, as they themselves write, because they use climate data from the capitals of coffee-producing countries, not the coffee-growing regions themselves.

From my own years of experience: When climate data from Managua, the capital of Nicaragua, is compared with that of Finca Santa Rita, in northern Nicaragua, the difference is huge. In Managua it's on average about 10 degrees warmer than at Santa Rita, there's no tree cover in the capital and it's significantly drier.

We admit that this is a crude assumption, because the climate near the capital is not necessarily representative for the climate in the areas in the country where coffee is grown, but global data on exact locations of coffee plantations are not easily obtainable.

In 2003, it was difficult to obtain exact coordinates of coffee farms. Geolocation was not yet a widely discussed topic in smallholder agriculture.

"Virtual Water"

Their globally weighted average resulted in 20,400 liters of virtual water per kilogram of roasted coffee. Virtual water is a concept that goes back to Tony Allan (1993). The idea: Every product carries an invisible amount of water in it – the water that was needed to produce it. If Switzerland imports wheat instead of growing it itself, it also "imports" the water that would have been needed for cultivation. Virtual water is therefore not water that is physically contained in the product, but purely a computational approach. The aforementioned Dutch scientist Hoekstra developed the "Water Footprint" concept from this.

How do you arrive at 140 liters? How did the authors calculate?

Step 1: How much water does a coffee plant need per year?

The authors use the CROPWAT model, which measures the evapotranspiration value of a coffee plant. In Brazil, they arrive at 12,000–13,000 m³ of water per hectare per year. That would be equivalent to a football field being about 1.8 meters underwater.

Step 2: How much coffee is harvested per hectare?

The authors used FAO average data from 1995–1999 from Brazil. They arrived at 1,100 kg of green coffee per hectare. Today we would assume at least double that. The 13,000 m³ divided by the 1,100 kg of green coffee per hectare equals 11,800 liters of water per kilogram of green coffee.

Step 3: From green coffee to roasted coffee

From 1 kg of green coffee, approximately 0.84 kg of roasted coffee results. The 11,800 liters divided by 840 g of roasted coffee equals 14,000 liters per kilogram. Globally weighted, it would be 20,400 liters per kilogram because in other countries significantly less yield per hectare is harvested.

Step 4: the 140 liters

These 20,400 liters divided by 7 g = 143 liters per cup. The 3-liter difference is negligible.

For a double espresso with 18g of coffee, it would then be 367 liters.

The core problem with the calculation

The first step in the calculation is decisive: The authors ask how much water a hectare planted with coffee plants evaporates per year, and arrive at a value of 13,000 m³. This is physically correct. But, and here lies the problem that so many simply jumped on: This figure does not distinguish where this water comes from.

The relevant question would be: How much additional, scarce water is demanded by coffee cultivation that would otherwise be available for other purposes?

Green, blue, and gray water

The water footprint concept distinguishes three components that have fundamentally different meanings for coffee.

Green water:

the rainwater that is stored in the soil and evaporates through the plant. For coffee, this should be over 95%, so almost the 140 liters. This water falls as rain on the land whether or not coffee grows there.

Blue water:

is water extracted from water bodies or groundwater. That's what we understand as "water consumption". For coffee, this is only about 1% of the water footprint (Revolve, 2023, which goes back to Mekonnen and Hoekstra 2011). Today it might be more, as the share of farms that irrigate artificially has increased.

Gray water:

is a theoretical construct. It refers to the amount of water that would be needed to dilute pollution caused by fertilizers to acceptable limit values. Particularly here, very precise attention must be paid. Studies from 2011 (Mekonnen and Hoekstra) and 2021 (Leal-Echeverri) arrive at a fivefold difference, solely through methodological assumptions.

The fundamental problem is that all three types of water are added together. More than an apples-and-oranges comparison, this calculation limps because it also brings in a theoretical construct.

Criticism of the concept

There has been much criticism from the scientific side and from several directions. Water footprints are not scientifically tested at all (Wichelns 2010, 2011) and water scarcity is not a global phenomenon. Precisely with coffee, this must really be viewed critically: Coffee is typically planted where rain falls. As climate change alters weather patterns, coffee was never planted in dry areas because it simply wouldn't have grown.

Since the water footprint was originally developed for irrigated dry regions, not for rain-fed agriculture in humid tropics – that is, where most coffee grows – the method is very misleading (Batchelor 2022).

If we were to continue the calculation using the aforementioned method, we would arrive at enormous regional differences: The authors arrive at 49,000 liters per kilogram in Ghana, but only 6,000 in Vietnam. The reason lies in the yield per hectare: If more is produced on a hectare, then the water value is divided by the yield. Higher harvest, less water consumption per kilo, so the calculation goes.

Today's far broader calculation of CO₂ emissions per hectare also draws in yield per hectare. High yield, lower emissions, but always per unit, so per kilo. We ourselves measured this way at Apas.

Processing methods and agroforestry

Washed processed coffees require water during the process as the cherries are conveyed through channels, then pulped and then sprayed with water (ecopulper) or washed in channels. The water used is fresh water so that, if possible, it is sterile and causes no contamination.

Dry processed coffees only need fresh water when the cherries are floated in a water channel before drying – the cherries that float to the top are removed. They are less dense and could have a defect.

Likewise, it's hardly surprising that coffee in full sun needs more water because it's often irrigated artificially. This is in contrast to coffee production in partial shade or in an agroforestry system, where coffee plants are surrounded by shade trees.

Realistic water consumption

In 2003, the researchers had difficulties obtaining precise farm data. That's why they, and subsequent scientists in the following years, relied on databases. This led much more to assumptions than certainties. The 140 liters per cup of coffee certainly sounds catchy and is effective in media.

But:

how much water does it really take to grow coffee for a cup of coffee? I asked our friends at Apas (Brazil), Cima Coffee (Honduras), Chacra d'Dago (Peru), and Mount Sunzu (Zambia). In short: It's not 140 liters per cup of coffee.

Mount Sunzu – Zambia: 22 liters

Washed processed

Luca Costa from Mount Sunzu wrote to me:

"I've looked at our calculations for sizing our irrigation system. I think the most important comment right at the beginning: These are our figures for our farm and our processing, so a farm-specific example and not a general value for coffee in general."

He's right about that. That's exactly what it's about – making specific calculations, analyzing them, and adjusting as needed.

Cultivation/Irrigation: 568.5 liters per kg cherry. Processing: 0.9 liters per kg cherry. Green coffee: 3,410 liters per kg. With a roasting loss of 20%, Luca arrives at 32 liters of water per cup (7 g). That's 77% less than the study says. Luca calculates conservatively, so with rather high figures. He assumes the actual water consumption in cultivation is 30% lower, and would thus arrive at 22 liters per cup (−84% vs. 140 liters).

Yair Keidar, Cima Café, Honduras: < 1 liter

Washed processed

"Indeed, in Honduras very little if any coffee farms use irrigation, so this is not really a big factor in water usage." Water for irrigation is hardly or not at all used in Honduras. For processing, however, the vast majority of producers in Honduras use wet processing. The efficiency depends on the equipment available.

Interestingly, IHCAFE – the Honduran coffee institute – picks up the 140 liters as a figure again:

There is a "limit" of 140 liters of water per quintal (bag) that the IHCAFE supposedly enforces, and a producer risks a fine if they exceed this amount of water.

There are the 140 liters again. They are the vague and methodologically questionable basis for a law that could fine producers if they use more than 140 liters to process 69 kg of green coffee.

At Finca San José in Santa Bárbara for example, total farm yield is about 80–120 bags (69 kg bags) and they are usually using around 130 liters per bag of washed coffee.

So that would be 1.9 liters per kilo of green coffee. For 7 g of roasted coffee, it would then be consequently 17 ml. For comparison: The espresso machine itself requires around 60–80 ml of brewing water for a double shot. The process water from the entire wet processing is therefore less than a quarter of the water you pour directly into the machine.

Chacra d'Dago, Peru: < 1 liter

Washed processed

Fresh water is used at Chacra d'Dago only for processing the cherries when coffee is wet processed. The water in the floater tanks, where cherries are sorted, is cleaned and reused.

Based on our records and operational practices, water use can be estimated in a range of approximately 90,000 to 152,000 liters, which is about 0.8 liters of water per kg of cherry under an efficient water management system.

0.8 liters of water per kg of cherries is then about 35 ml per 7 g of roasted coffee, 99% less than 140 liters of water. The modern facility with a closed flotation circuit is very efficient and purifies the water, which would bring the value even lower.

Apas, Brazil: 0 liters

Dry processed

At Apas, the coffee plants are not artificially irrigated and the cherries are dry processed as naturals. Thus they consume no liters of fresh water. What was not included in all the calculations is the amount of fresh water for liquid fertilizer that is applied to the leaves. But that would be minimal volumes and wouldn't change the calculations.

For washed coffees from the region, according to Mauricio from Apas, 30 liters per 60 kg of green coffee would be needed for washing. Another 50 liters for pulping. Combined, that would then be 11.6 ml per cup, also 99.9% less than 140. That would then be 4.4 ml per cup of coffee.

What real water problems are

The 140-liter figure is no longer relevant today, but it persists stubbornly. Meanwhile, we need to talk about other water problems that really have an impact and endanger local ecosystems.

Irrigation in water-scarce regions

Close attention must be paid to where the water comes from when coffee is artificially irrigated. According to EMBRAPA, Brazil's agricultural research institute, from 2022 to 2024, 14% more coffee plantations in Brazil have been artificially irrigated. Everywhere high yields are sought (e.g., also Vietnam), water requirements must be carefully checked.

Wastewater from wet processing

Coffee wastewater contains tannins, phenols, and alkaloids that inhibit biodegradation. Oxygen is consumed and anaerobic conditions develop. Untreated wastewater is highly organically polluted, which has a lethal effect on river organisms and is not drinkable. I've visited farms that fed the agua miel, the wastewater from washed production, unfiltered into the fields. I've also seen farms that had state-of-the-art water treatment facilities and actively addressed the problem.

Climate change and altered rainfall patterns

However, the biggest water problem for coffee is still to come: Climate Central documents for 2026 an average of 47 additional days per year with temperatures above 30°C in the 25 most important growing countries. That's when shade, water storage, and artificial irrigation become increasingly important.