How to Use a French Press: Ratio, Grind, Time and the Sludge Question

The French press has the best marketing in coffee: one glass cylinder, one plunger, no paper, no pump, no app. It also has the worst reputation for sludge. Both are earned. Here's how to make a good pot, what the science actually says about steep time and grind, where the "French" part of the story falls apart, and the one thing about unfiltered coffee that's worth knowing if you drink a lot of it.

The short version

  • Ratio is the main dial. In full-immersion brewing, the coffee-to-water ratio sets the strength. Grind, roast and water temperature (80–99°C) barely move where the brew ends up.[1]
  • A solid starting recipe: about 55 g of coffee per liter of water (roughly 1:18), water just off the boil, coarse grind, about 4 minutes, then press and pour.[2][3]
  • It doesn't keep getting "more extracted" forever. Extraction climbs, then levels off. Lab brews plateaued at around 21%.[1]
  • The mesh lets oils and fine particles through. That's the body people love, and it's also why press coffee carries cholesterol-raising cafestol that paper filters mostly catch.[5][6][7]
  • The "French" part is messy. There's an 1852 Paris patent, but the sealing plunger that made it work was patented by Milanese inventors from 1928 on.[9][10][11]

How a French press actually brews

A press is a full-immersion brewer. All the water and all the coffee sit together for the whole brew, then a mesh disc on a rod pushes the grounds to the bottom. Drip and pour-over are the opposite: fresh water keeps flowing through a bed of grounds and out through a filter.

That difference matters. In 2021, UC Davis researchers measured full-immersion brews across a wide range of ratios, grind sizes, roasts and temperatures. The strength of the brew (total dissolved solids, or TDS) was roughly inversely proportional to the water-to-coffee ratio. The extraction yield, the share of the coffee that ended up dissolved, came out at about 21% almost no matter what. The equilibrium barely changed with grind size, roast level, or brew temperature between 80 and 99°C. Their conclusion, in plain English: in immersion brewing, pay attention to your ratio, because it's the most important setting you have.[1]

Translation for your kitchen: if your press coffee tastes weak, use more coffee. If it tastes like a punishment, use less. Fiddling with water temperature by a few degrees isn't going to rescue a bad ratio.

A starting recipe (adjust to taste)

The Specialty Coffee Association's professional tasting method, called cupping, is itself an immersion brew. It uses 8.25 g of coffee per 150 ml of water, which is 0.055 g per ml, with water at 200°F ±2°F (92.2–94.4°C).[2] The SCA's Golden Cup guidance for brewed coffee lands at about the same ratio, 55 g per liter ±10%.[3] That makes 55 g/L a sensible place to start in a press:

  • 350 ml press (one big mug): about 19 g coffee
  • 500 ml press: about 27–28 g coffee
  • 1 liter press (the "8-cup"): about 55 g coffee
  1. Grind coarse. Think coarse sand or breadcrumbs, not powder. Finer grinds make more particles small enough to slip past the mesh. (If you buy pre-ground, pick a coarse grind.)
  2. Heat the water to just off the boil, around 92–94°C. If you have no thermometer, let a boiled kettle sit for about 30 seconds. Given the research above, don't lose sleep over a couple of degrees.[1]
  3. Add coffee, then all the water. Stir once or twice so every bit of coffee is wet.
  4. Wait about 4 minutes. That's tradition rather than law. The cupping ritual breaks the crust at four minutes, and plenty of brewers simply borrow the number.[1] Go longer if you want it stronger.
  5. Press slowly and steadily. If it fights back hard, your grind is probably too fine.
  6. Pour it all out into cups or a carafe. Don't leave it sitting on the grounds for an hour of reheats.
  7. Leave the last spoonful in the pot. The final pour is where the settled silt lives.

Myth check: "If you steep it too long, it gets endlessly bitter"

Not quite. In the UC Davis work, TDS in immersion brews rose and then plateaued. In their one-liter beakers that took around 20 minutes. Hotter water got there faster, but it didn't change where it ended up.[1] A follow-up sensory study from the same lab found that roast level had the biggest effect on flavor, then brewing temperature. Brew time, especially the long times after TDS had plateaued, had "subtler impacts than expected." Sweetness went down as TDS went up.[4]

So a 4-minute press is actually pulled before the plateau. Going to 6 or 8 minutes makes a stronger, less sweet cup, not a chemically ruined one. The real reasons to pour promptly are boring: the coffee cools and goes stale in the pot, and the grounds at the bottom keep releasing silt every time you tilt it.

The sludge and the cholesterol bit

The metal mesh is the press's whole personality. It lets through coffee oils and very fine particles that a paper filter would trap, and that's where the heavier body comes from. It also lets through two compounds in coffee oil, cafestol and kahweol, that raise blood cholesterol.

  • Wageningen University research led by Rob Urgert and Martijn Katan measured French press coffee at about 3.5 mg of cafestol per 150 ml cup. Boiled coffee came in at about 3.0 mg and Turkish/Greek at about 4.7 mg, while drip-filtered, percolated and instant brews had negligible amounts. Urgert estimated that five cups a day of French press, boiled or Turkish coffee could raise serum cholesterol by about 0.2–0.3 mmol/L.[5]
  • The fines count too. French press coffee contained about 1.5 g of floating particles per liter (Turkish and boiled coffee had 2–5 g). In a feeding trial, eating 8 g of fine grounds a day for three weeks raised cholesterol.[6]
  • A 2025 Swedish study measured a French press brew at about 87 mg of cafestol per liter. The home paper-filtered drip coffees came in at a median of about 12 mg/L.[7] Numbers vary a lot with recipe and coffee, so read these as orders of magnitude, not label values.
  • Roast and method shift how much cafestol gets out of the grounds. In one lab comparison, French press and boiled brews made with the lightest roast had the highest cafestol extraction yields.[8]

What this does not mean: that a French press is dangerous, or that you should panic about one cup. It means that if you drink several cups a day and your LDL is something you and your doctor are watching, paper-filtered coffee is the evidence-backed choice. We haven't found a study testing "press it, then pour it through paper," so we won't pretend that trick is proven.

So is it actually French? A short, contested history

The folklore: an old Frenchman boils water for his coffee, forgets the grounds, and they float. A passing merchant (often Italian, depending on who's telling it) sells him a metal screen, he pushes it down with a stick, and the two of them get rich. It's a lovely story, and nobody has documented it. Coffee historians treat it as an invention.[9][10]

The 1852 patent: on 27 March 1852, Henri-Otto Mayer and Jacques-Victor Delforge filed a Paris patent for a cafetière à pression et à filtrage instantanés, a "press coffee pot with instant filtering." Its filter was cloth on a plunger. Historians who have examined the patent note that it had no real seal against the pot walls, and there's no evidence it was ever made or sold.[9][10] (We couldn't open the original record in France's INPI patent archive, so this part rests on those secondary accounts.)

The Milan fix: the press as we know it shows up in Italian patents. US patent 1,797,672 was filed on 2 April 1929, with an Italian priority date of 2 April 1928, and granted in 1931. It names Ugo Paolini of Milan as inventor, assigning half to Attilio Calimani and half to Giulio Moneta. It describes a filtering "piston-like member," a rod through the lid, and a spring that holds the piston wherever you leave it.[11] In 1934 a Milan patent by Bruno Cassol, assigned to Gemma Barelli Moneta, described a sliding filter for infusions.[12] Coffee historians credit that Moneta/Cassol design with the double-disc-and-gauze filter that modern presses still copy.[9]

The Chambord: Faliero Bondanini filed US patent 2,900,896 for a "coffee filter pot" in 1957, with a Swiss priority date of 1955. It was granted in 1959.[13] According to a US federal appeals court, a French company, Martin S.A., distributed a successful press called the Chambord from the mid-1950s until 1991, when Bodum bought Martin. Bodum companies have sold the Chambord design and name ever since.[14] (Popular histories often say "1958" for the Chambord. The patent paperwork points to the mid-to-late 1950s, so treat any single year loosely.)

So the honest answer is: it was French-patented first, Italian-engineered into something that worked, and made famous by a French factory and then a Danish brand. "French press" is the American name. In France it's a cafetière à piston, and in Britain it's usually a "cafetière."[10][14]

What to put in it

Because immersion keeps body and oils, presses suit coffees you'd describe as rich or chocolatey. Brighter coffees work too, with a little more texture than in a pour-over. If you buy pre-ground, choose a coarse grind.

  • French Roast: our dark roast blend from Central America. Yes, French Roast in a French press. We know.
  • Colombia: a medium roast with tasting notes of dried orange, berry and chocolate.
  • Ethiopia Natural: a medium roast, natural process, fruit-forward.
  • Peru Decaf: Swiss Water Process decaf for evening pots.

Or browse everything on the shop page.

Sources

  1. Jiexin Liang, Ka Chun Chan and William D. Ristenpart, "An equilibrium desorption model for the strength and extraction yield of full immersion brewed coffee," Scientific Reports 11, 6904 (2021): nature.com
  2. Specialty Coffee Association, SCA Protocols & Best Practices (revised 2018), Cupping Standards 3.1 (coffee-to-water ratio) and 3.3 (water temperature); copy hosted by a third party: SCA Protocols & Best Practices (PDF)
  3. Specialty Coffee Association, Golden Cup Standard and Brewing Best Practices (55 g/L ±10%), in the same document: SCA Protocols & Best Practices (PDF)
  4. Jiexin Liang, Mackenzie E. Batali, Catherine Routt, William D. Ristenpart and Jean-Xavier Guinard, "Sensory analysis of the flavor profile of full immersion hot, room temperature, and cold brewed coffee over time," Scientific Reports (published 20 August 2024): nature.com
  5. R. Urgert, G. van der Weg, A. G. M. Schulz and M. B. Katan, "Cafestol and kahweol in coffee brews: effects on serum cholesterol in humans" (abstract), Wageningen University & Research portal: research.wur.nl
  6. R. Urgert, A. G. Schulz and M. B. Katan, "Effects of cafestol and kahweol from coffee grounds on serum lipids and serum liver enzymes in humans," American Journal of Clinical Nutrition 61(1):149–154 (1995): PubMed
  7. E. Orrje, R. Fristedt, F. Rosqvist, R. Landberg and D. Iggman, "Cafestol and kahweol concentrations in workplace machine coffee compared with conventional brewing methods," Nutrition, Metabolism and Cardiovascular Diseases (2025): PubMed
  8. C. Zhang, R. Linforth and I. D. Fisk, "Cafestol extraction yield from different coffee brew mechanisms," Food Research International 49(1):27–31 (2012): doi.org
  9. Just Scoof, "The cafetiere story: A new history of the cafetiere or French Press" (patent-based history, including the 1852 Mayer and Delforge and 1934 Moneta/Cassol patents): justscoof.com
  10. Retrowow, "History of the cafetière": retrowow.co.uk
  11. U. Paolini (assignor of one-half to Attilio Calimani and one-half to Giulio Moneta), US Patent 1,797,672, "Apparatus for preparing infusions, particularly for preparing coffee," filed 2 April 1929, granted 24 March 1931: Google Patents
  12. B. Cassol (assignor to Gemma Barelli Moneta), US Patent 2,053,021, "Apparatus for filtration of infusions," Italian priority 2 March 1934, granted 1 September 1936: Google Patents
  13. F. Bondanini, US Patent 2,900,896, "Coffee filter pot," filed 16 July 1957, granted 25 August 1959: Google Patents
  14. Bodum USA, Inc. v. La Cafetière, Inc., No. 09-1892 (7th Cir., decided 2 September 2010): FindLaw