Use weight rather than spoons. A tablespoon of coffee varies enormously with grind size and bean density, and it is why the same recipe gives different results on different days.
Filter and pour-over
The standard is roughly 60 g (2.1 oz) of coffee per liter of water, which is a 1:16.7 ratio. In practice:
- 250 ml (8.5 fl oz) cup — 15 g (0.5 oz) of coffee
- 12 fl oz (355 ml) mug — about 21 g (0.75 oz)
- 500 ml — 30 g (1.1 oz)
- 1 liter batch — 60 g (2.1 oz)
Ratios between 1:15 and 1:18 are all reasonable. Tighter (1:15) gives a stronger, heavier cup; wider (1:18) gives a lighter, more tea-like one. Adjust to taste and keep it consistent while you dial in the grind.
French press
The same ballpark, often a touch stronger: 60 to 70 g (2.1 to 2.5 oz) per liter, coarse grind, four minutes.
Moka pot
Fill the basket level and untamped, and fill the base to the valve. The pot’s geometry sets the ratio — there is nothing to weigh, and you should brew the size the pot is rated for.
Espresso
A completely different measure. 18 g (0.6 oz) of coffee producing 36 g (1.3 oz) of espresso, weighed rather than eyeballed, in 25 to 32 seconds. That is a 1:2 ratio, and it is by weight of liquid rather than volume, because crema volume varies wildly.
Why weighing beats scooping
Fifteen grams of a light roast occupies noticeably more space than fifteen grams of a dark one, so a scoop delivers different amounts of coffee depending on what is in the bag. A $20 scale removes that variable permanently.
Strength versus extraction
If your coffee is weak, adding more coffee makes it stronger. If it is sour, that is under-extraction and more coffee will not fix it — grind finer instead.
What people want to know
How many grams in a tablespoon of coffee?
Roughly 5 to 7 g (0.2 to 0.2 oz) depending on grind and roast, which is exactly why the measure is unreliable. Use a scale.
Should I use more coffee for a bigger mug?
Yes, proportionally. Brewing 250 ml of coffee and topping it up with hot water gives a different, weaker drink than brewing 350 ml properly.
Last reviewed 30 October 2026