Coffee-to-Water Ratio: A Reference for Every Brew Method

For filter coffee, use 55 to 60 grams of coffee per litre of water — 1:18 to 1:16.7 by weight, or 28 to 30 g for a 500 ml brew. If you want one number to start from, take 60 g/L and adjust from there.

The ratio is the easiest brewing variable to get right, which is why it attracts more attention than it deserves. It is also the one most guides stop at, and stopping there is why so many people own a scale and still cannot fix a bad cup.

Ratios by method

Espresso belongs on a different scale entirely, for reasons worth a section of its own.

What the ratio controls, and what it does not

Two independent things decide how coffee tastes, and conflating them is the single most common brewing mistake.

Strength is how much dissolved coffee is in the water — measured as TDS, total dissolved solids. The brew ratio sets it. More coffee per litre, stronger cup.

Extraction is what fraction of the coffee's mass you dissolved, measured as yield. Grind size, contact time and water temperature set it. Low extraction tastes sour, thin and salty; high extraction tastes bitter, hollow and drying.

A cup can be strong and under-extracted, or weak and over-extracted. These are different faults with opposite fixes, and the ratio only addresses one of them. Watering down a bitter coffee makes a weaker bitter coffee: the harsh compounds are already dissolved, and dilution cannot un-dissolve them. Grind coarser instead.

This is also why "use less coffee" is bad advice for a cup that tastes burnt, and why "use more coffee" rarely fixes one that tastes sour.

Where 55–60 g/L came from

In 1952 the Coffee Brewing Institute — funded by the National Coffee Association and the Pan-American Coffee Bureau — set out to answer what makes coffee taste good. Ernest Earl Lockhart, a biochemist at MIT and the Institute's first director, measured dissolved solids in brewed coffee with a refractometer and matched the readings against what taste panels preferred.

His 1957 report gave the industry two numbers that have barely moved since: a strength of 1.15–1.35% TDS and an extraction yield of 18–22%. He plotted them as the Brewing Control Chart, the strength-versus-extraction grid still printed on café walls, with a rectangle in the middle marked ideal. The Specialty Coffee Association inherited those figures as the Golden Cup standard, and the brewing ratio band of 55–60 g/L falls out of the arithmetic: it is the ratio that lands you in the rectangle when extraction goes to plan.

Worth knowing, because it is usually left out: those numbers describe what a particular group of mid-century American drinkers liked, in an era of darker commodity blends. It was real research, not folklore. It was also one population, one coffee style, one moment.

The 2023 revision: there is no single ideal zone

In 2023 a team from the UC Davis Coffee Center — Jean-Xavier Guinard, Scott Frost, Mackenzie Batali, Andrew Cotter, Lik X. Lim and William Ristenpart — published new Brewing Control Charts in the Journal of Food Science, built from calibrated sensory panels and consumer preference testing across the same TDS and extraction ranges Lockhart used.

Two findings matter for anyone choosing a ratio.

First, they mapped specific flavours onto the grid — bitter, sour, citrus, roasted, viscous, astringent, sweet, burnt, black tea, berry, dried fruit — rather than the handful of labels on the old chart. You can now read the chart as a prediction of what a cup will taste like, not just whether it is inside a box.

Second, and more disruptive: consumer preference did not form one cluster. It formed two, with different likes and dislikes across the same ranges. The old chart asserted a single ideal zone. The new one shows that good coffee means materially different things to different groups of drinkers.

The practical consequence is not that the numbers are wrong. 60 g/L is still an excellent place to start. It is that the rectangle was never a target so much as a description of one audience's taste, and you are allowed to sit outside it on purpose.

The arithmetic nobody shows you

Here is where most ratio guides quietly mislead. The ratio you set is coffee to water added. It is not coffee to coffee in the cup, because the grounds keep a large share of the water.

Wet coffee grounds retain roughly 2 g of water per gram of dry coffee. So:

• Brew 15 g of coffee with 250 g of water. The spent bed holds about 30 g. You get roughly 220 g in the cup, not 250 g.
• Brew 30 g with 500 g. The bed holds about 60 g, leaving about 440 g.

That gap is why a 1:16.7 brew tastes stronger than the ratio implies — the coffee that dissolved is concentrated into less liquid than you poured. It is also why extraction yield is calculated from the beverage, not from the water added:

Extraction yield = (beverage mass × TDS%) ÷ dose

Two worked examples, using realistic numbers:

• Pour-over. 15 g dose, 250 g water, about 220 g in the cup, measured at 1.35% TDS. Dissolved solids: 220 × 0.0135 = 2.97 g. Yield: 2.97 ÷ 15 = 19.8%. Inside the classic window, at the strong end.
• French press. 30 g dose, 500 g water, about 440 g in the cup at 1.25% TDS. Dissolved solids: 440 × 0.0125 = 5.5 g. Yield: 5.5 ÷ 30 = 18.3%. Same ballpark, noticeably lighter in the mouth.

You do not need a refractometer to brew well. But knowing the formula explains the arithmetic behind advice that otherwise sounds arbitrary — including why immersion brewers feel weaker than a pour-over at an identical ratio. They lose more water to a bed that never drains.

Why the methods differ

Percolation brewers — V60, Chemex, Kalita, batch machines — pass fresh water through a bed and out. Water keeps arriving unsaturated, which extracts efficiently, so these methods sit comfortably at 55–60 g/L with a medium-fine grind.

Immersion brewers — French press, cupping bowls, steeped cold brew — sit the coffee in a fixed volume of water. As solubles dissolve, the water gets closer to saturation and extraction slows down on its own. That self-limiting behaviour makes immersion forgiving of time but a little less efficient, so the same ratio gives a softer, lighter cup. Keep the ratio, grind coarser, and let time do the work.

AeroPress is both, and its ratio range is wide because people use it in opposite ways: a short, strong, espresso-adjacent brew at 1:12, or a long filter-style brew at 1:16. Neither is wrong; they are different drinks.

Moka pot is set by the hardware. The basket holds what it holds and the boiler takes what it takes, which lands around 1:7 to 1:10 — far stronger than filter, because it is meant to be drunk in small volumes or cut with milk. Fill the basket level, do not tamp, and take it off the heat as soon as it starts to sputter.

Cold brew needs the retained-water arithmetic more than any other method. At 1:5, a 100 g dose in 500 g of water gives up about 200 g to the grounds. You pour 500 g and recover roughly 300 g of concentrate — two-fifths of your water stays behind in the bag. Budget for that before you plan a batch, then dilute the concentrate 1:1 or further to taste.

Espresso is a ratio of something else

An espresso ratio compares dose to liquid in the cup, not dose to water pushed through. A standard modern shot is 1:2 — 18 g of coffee in, 36 g of espresso out, usually in 25 to 30 seconds. A ristretto is shorter, around 1:1.5; a lungo is longer, 1:3 or beyond.

The machine actually sends more water than that through the puck, because the puck keeps a lot of it. A spent 18 g puck comes out weighing roughly twice what went in, so a 36 g shot has seen closer to 55–60 g of water. This is why 1:2 and 1:16 are not comparable numbers, and why putting them in the same table without explanation confuses people learning both.

The 2020 Matter paper by Cameron, Hendon and colleagues is worth knowing here. Modelling and experiment together showed that grinding very fine promotes uneven flow through the puck — channelling — which hurts reproducibility and wastes coffee. Their counterintuitive recommendation was to use less coffee ground coarser: around 15 g rather than 20 g, with a shorter ratio, for shots that are easier to repeat. If your espresso is inconsistent shot to shot, that is a more promising lever than another 0.5 g of dose. The dialing-in guide walks through the whole process.

Changing the ratio on purpose

Change one variable at a time, and taste before you decide which one.

Note which rows move the grind and which move the ratio. Sourness and bitterness are extraction problems, and worth diagnosing properly before you change anything. Faint and heavy are strength problems. Diagnose which one you have before reaching for the scale.

When you scale a recipe, keep the ratio and expect the grind to need work. Doubling a pour-over doubles the depth of the bed, which slows drawdown and raises extraction, so the bigger brew usually tastes heavier at the same setting. Coarsen a little. Ratios scale linearly; brews do not.

When to leave the range

A few cases where 55–60 g/L is the wrong target, and knowing why is the point:

• Very light roasts. Dense, lightly roasted coffee resists extraction, and plenty of them taste sweetest above 22% yield — past the old chart's edge. Grind finer, brew hotter, and judge by taste rather than by the rectangle.
• Iced coffee brewed onto ice. Brew with less hot water and let a measured amount of ice make up the rest: about 60–70 g of coffee per litre of total water, ice included, with a third to two-fifths of it as ice. The hot portion brews strong and the melt brings it back to normal strength.
• Milk drinks. Milk flattens acidity and masks strength, which is why espresso and moka exist at the ratios they do. A filter coffee intended for a large milky drink needs to be brewed stronger than one you drink black.
• Water you cannot change. Very soft water extracts poorly and heavily buffered water flattens acidity. If your tap makes every coffee taste flat regardless of ratio, the water is the variable, not the recipe.

The useful posture is the one the new research points at: treat 60 g/L as a well-evidenced default rather than a rule, know which knob fixes which fault, and let the cup settle arguments.