Independent brewing lab — not affiliated with PourX, Abilliant or the domain's previous owner.
pourxlabthe pour-over lab

Reference · filters

Coffee filters: paper vs metal vs cloth

Every filter holds back the grounds. What separates paper, metal mesh and cloth is what else gets through — fine particles and coffee oil, and with the oil the diterpenes cafestol and kahweol. Below: the three materials side by side, what the cafestol studies actually measured (and where they disagree), bleached vs unbleached paper, sizes, rinsing, and the 1908 filter that started paper brewing.

Updated 2026-10-26 · sources listed at the end

What paper, metal mesh and cloth filters hold back, and where the cafestol went in a paper-filter studyPanel A shows three filters side by side, each with coffee grounds, fine particles and oil droplets above it and a cup below. All three hold back the grounds. Below the paper filter the cup holds only liquid with two tiny particles. Below the metal mesh the cup holds fine particles and oil droplets. Below the cloth the cup holds a few oil droplets. Panel B is a to-scale bar from a 2017 paper-filter study: 87.45 percent of the cafestol stayed in the spent grounds, 12.41 percent was retained by the paper, and 0.15 percent reached the brew. Numbered markers 1 to 8 match the list below. A · What reaches the cup PaperMetal meshCloth 1 2 3 4 5 6 7 8 B · Where the cafestol went spent grounds 87.45%paper 12.41%brew 0.15% bar drawn to scale, 100% = cafestol in the roasted coffee
  1. Coffee grounds: held back by all three filters.
  2. Fines: the smallest particles. Paper lets through only particles smaller than roughly 10–15 µm; with a metal mesh, finer grounds can seep through or around the filter.
  3. Coffee oil: carries the diterpenes cafestol and kahweol.
  4. Paper: about 100 g/m² filter paper with fibres about 20 µm wide; removes the oily diterpenes.
  5. Metal (or nylon) mesh: does not remove the diterpenes, so oil and some fines reach the cup.
  6. Cloth: lets more of the oil through than paper does.
  7. Brew: drop counts are illustrative, not measured amounts.
  8. Cafestol balance in one paper-filter study (Rendón et al. 2017): most of it never left the grounds.
Drawn by pourxlab.com for this page. Panel A follows the Wikipedia coffee filter and French press articles (sources 13 and 15); not to scale, particle and droplet sizes illustrative. Panel B is to scale from Rendón, Scholz and Bragagnolo 2017 (source 1).

Paper, metal and cloth compared

FilterWhat gets throughIn the cup, as sources describe itCafestol, as studies report itReuse and care
Paper
about 100 g/m², bleached or unbleached		Particles under roughly 10–15 µm; the oily diterpenes are removedSaid to taste "somewhat fruitier" than permanent filters; some baristas report a "paperish" taste0.15% of the coffee's cafestol reached the brew (Rendón 2017); home paper-filtered median 12 mg/L (Orrje 2025)Single use; a filter that is not strong enough can tear and let grounds into the pot
Metal or nylon mesh
stainless steel wire or nylon		Oil and diterpenes; finer grounds can seep through or around the meshMore oil and fine sediment than paperFrench press 86.8 and percolator 91.2 mg/L; metal-filter workplace brewing machines median 176 mg/L (Orrje 2025)Reusable. The Orrje authors suggest, as a hypothesis, that cleaning could make a metal filter more permeable
Cloth
cotton, linen, flannel, muslin, hemp and others		More oil than paperLower solids content in one study (Wuerges 2020)Boiled coffee 939 mg/L, 28 mg/L once poured through fabric (Orrje 2025); more diterpenes per gram of solids than paper (Wuerges 2020)One maker's instructions: wash, store wet in a little water in the fridge, never let it dry out

Read the cafestol column within each study only. The papers measure different things (mg per litre, mg per 50 ml serving, percent of the bean's cafestol) with different coffees, so their absolute numbers do not line up — see the disagreements below.

What the cafestol studies measured

Cafestol makes up about 0.4–0.7% of an arabica bean by weight (Wikipedia); a Nottingham study measured an average of 603 mg per 100 g of roast and ground coffee. How much of it ends up in the cup depends on the brew and the filter:

StudyWhat was comparedMain finding
Zhang, Linforth & Fisk 2012Boiled, Turkish, French press and moka pot, by roastFrench press, Turkish and boiled had the highest cafestol; extraction yield peaked at 6.5% (light roast, French press)
Rendón, Scholz & Bragagnolo 2017Where cafestol goes in paper-filter brewing87.45% stayed in the spent grounds, 12.41% on the paper, 0.15% in the brew; hot water extracts little of the lipid fraction
Wuerges et al. 2020Cloth- and paper-filtered, espresso, instantLow diterpenes in all brews; per 50 ml, filtered and espresso similar; cloth higher per gram of solids; instant had no detectable cafestol
Orrje et al. 2025Swedish workplace machines vs home brewsMetal-filter brewing machines between paper-filtered and unfiltered coffee; boiled highest among non-espresso brews; some espresso up to 2447 mg/L

Where the studies disagree

  • Is it the paper or the grounds? Wikipedia credits the paper with retaining the diterpenes. Rendón and colleagues found the paper held about an eighth of the cafestol; most stayed in the grounds because hot water dissolves little of the oil.
  • Espresso. Harvard's Nutrition Source says espresso has "moderate amounts" of diterpenes. Wuerges et al. found espresso and paper-filtered brews similar per 50 ml, while Orrje et al. measured a median of 1059.3 mg/L in their espresso samples.
  • Absolute levels. Wuerges et al. report 0.11–0.14 mg cafestol per 50 ml (none detected in instant); Orrje et al. report a 12 mg/L median for home paper-filtered coffee — a several-fold gap that reflects different coffees, methods and units.

What the health sources say

This section summarises published research and official guidance; it is not medical advice.

  • Harvard Nutrition Source: unfiltered coffee such as French press and Turkish contains diterpenes that "can raise bad LDL cholesterol and triglycerides"; filtered (drip) and instant coffee "contain almost no diterpenes".
  • Urgert et al. 1996 (BMJ): in a 24-week randomised trial, 46 healthy adults drank five to six strong cups (0.9 L) a day. Cafetière coffee raised LDL cholesterol by 9–14% at first and the liver enzyme ALT relative to filtered coffee; "all increases were reversible" after the trial.
  • Urgert et al. 1997 (AJCN): 61–64 mg a day of pure cafestol for 28 days raised LDL by 0.57 mmol/L in 10 men; kahweol added little to the lipid effect, so unfiltered coffee's cholesterol-raising potential depends mainly on cafestol.
  • Tverdal et al. 2020: among 508,747 Norwegians aged 20–79, filtered coffee was associated with lower mortality than no coffee; nine or more cups a day of unfiltered coffee, compared with one to four of filtered, was associated with higher heart-disease mortality in men (HR 1.21). Intake came from questionnaires; these are associations, not proof of cause.
  • Orrje et al. 2025: the authors conclude that "thoroughly filtered coffee seems like the preferable choice for cardiovascular health," including at work.

Bleached vs unbleached paper

Both white (bleached) and brown (unbleached) paper filters are made. Early work identified a chlorine bleaching process as the source of dioxins in bleached paper. A 1994 FDA notice cites a 1988 industry study in which 65 to 90 percent of the dioxin equivalents in bleached coffee filters could migrate into the coffee, and records that manufacturing changes significantly reduced them. FDA tentatively concluded that bleached food-contact paper is safe. A 2010 TU Wien thesis notes the industry has moved from molecular chlorine to chlorine dioxide and to oxygen and peroxide ("total chlorine free") bleaching; after testing filters sold in Austria for migrating substances, it concluded they "all seem to be safe." Neither source ranks white against brown paper on taste.

Shapes and sizes

ShapeSizes as published
Cone (e.g. V60)HARIO bag sizes 01 (1 cup), 02 (1–4 cups), 03 (1–6 cups)
Fan or boat (US numbering)#0, #1, #2, #4, #6, similar to the German sizes 100, 101, 102, 1×4 and 1×6; #2, #4 and #6 are the most popular
Basket8–12 cup home size, larger restaurant sizes
Round disk94 mm (the 1908 original), 120 mm; about 63 mm for the AeroPress
Half-moon or squareChemex, paper 20–30% thicker than regular filters

The numbers are not interchangeable across systems: Wikipedia notes that the US fan-shaped "#1" must not be confused with Melitta's round "1" disk or HARIO's 01 cone. Recipes for these brewers: V60, Chemex, AeroPress.

Rinsing

Some baristas report a "paperish" taste from paper filters and recommend flushing the filter with hot water before adding the coffee. Metal and porcelain holders store more heat than glass or plastic, so pre-heat them to avoid a large temperature drop during the pour. The rinse step is part of the pour-over method.

Where the paper filter came from

According to the German Patent and Trade Mark Office, Melitta Bentz of Dresden, a mother of three, tested a perforated brass cup with blotting paper from her son's exercise book. She registered a "coffee filter with a domed underside, recessed bottom and inclined flow holes" on 20 June 1908; it was granted on 8 July 1908. Her company entered the commercial register on 15 December 1908 with 72 Reichspfennig of capital, starting with 100 cartons of filter paper and 50 filters; more than 1,200 sold at the 1909 Leipzig Trade Fair. The paper filter in its familiar shape was patented in the 1930s (DE 653796A, 1937), and the company, now in Minden, makes 50 million filters a day.

Filter checklist

  • Least oil and sediment: paper. More oil and body: metal mesh or cloth.
  • Match the shape and size system to the brewer (01/02/03, #2/#4/#6, basket, disk).
  • Flush paper with hot water before adding coffee; pre-heat metal or porcelain holders.
  • Cloth: wash after use and keep it wet, per the maker's instructions.
  • Comparing brews by cafestol? Use one study's numbers, not a mix.
  • Health questions: the studies above are summaries, not medical advice.

Sources

  1. Rendón M.Y., dos Santos Scholz M.B., Bragagnolo N. (2017) Is cafestol retained on the paper filter in the preparation of filter coffee? Food Research International 100:798–803 (FAO AGRIS record; 87.45% / 12.41% / 0.15%)
  2. Zhang C., Linforth R.S.T., Fisk I.D. (2012) Cafestol extraction yield from different coffee brew mechanisms. Food Research International 49(1) (University of Nottingham repository)
  3. Wuerges K.L., Dias R.C.E., Viegas M.C., Benassi M.T. (2020) Kahweol and cafestol in coffee brews: comparison of preparation methods. Revista Ciência Agronômica 51(1) (DOAJ record; cloth, paper, espresso, instant)
  4. Orrje E., Fristedt R., Rosqvist F., Landberg R., Iggman D. (2025) Cafestol and kahweol concentrations in workplace machine coffee compared with conventional brewing methods. Nutrition, Metabolism and Cardiovascular Diseases 35:103933 (Chalmers research repository, record and full text)
  5. Urgert R. et al. (1996) Comparison of effect of cafetière and filtered coffee on serum concentrations of liver aminotransferases and lipids: six month randomised controlled trial. BMJ 313:1362–1366 (PubMed Central)
  6. Urgert R., Essed N., van der Weg G., Kosmeijer-Schuil T.G., Katan M.B. (1997) Separate effects of the coffee diterpenes cafestol and kahweol on serum lipids and liver aminotransferases. American Journal of Clinical Nutrition 65(2):519–524 (FAO AGRIS record)
  7. PACE-CME (2020) Drinking filtered coffee associated with lower mortality — report on Tverdal A., Selmer R., Cohen J.M. et al., European Journal of Preventive Cardiology 2020, doi:10.1177/2047487320914443
  8. Harvard T.H. Chan School of Public Health — The Nutrition Source: Coffee (diterpenes in unfiltered, espresso, filtered and instant coffee)
  9. U.S. Food and Drug Administration — Federal Register 59(70), April 12, 1994, Docket 93N-0352: dioxins and furans in bleached food-contact paper (coffee filter migration, risk estimates)
  10. Fellner M. (2010) Migration from coffee filter papers. Diploma thesis, TU Wien (record and full-text PDF; bleaching methods, migration tests)
  11. German Patent and Trade Mark Office (DPMA) — The invention of the coffee filter (Melitta Bentz, 1908)
  12. HARIO UK — cloth filter for the Woodneck drip pot (maker's care instructions; cited for care only)
  13. Wikipedia — Coffee filter (secondary: paper grammage and pore size, diterpenes, cloth, shapes and size numbers, rinsing, history)
  14. Wikipedia — Cafestol (secondary: 0.4–0.7% of arabica beans, unfiltered vs paper-filtered drinks)
  15. Wikipedia — French press (secondary: stainless steel or nylon mesh, fines seeping through)