Almost everything written about coffee filters is a taste preference. One difference is not, and it is the reason to read this page: a paper filter holds back cafestol and kahweol, the coffee oils that raise LDL cholesterol. A metal filter lets them through. A cloth filter sits in between and, on the measurements below, sits much closer to paper than to metal. If your cholesterol is something you or your doctor are watching, that is the whole decision. If it is not, pick whichever filter gives you the cup you like — for flavour, paper is the cleanest, metal the heaviest, cloth in the middle.
Table of Contents
The cholesterol finding, checked
An earlier internal review of this site flagged the cafestol claim as unverified, so it has been checked from the sources rather than inherited. Two hold it up.
The effect. A toxicological review prepared for the National Institute of Environmental Health Sciences states it directly: “The cholesterol-raising effect of boiled coffee in humans has been linked with these diterpenes. Studies have shown that an intake of cafestol and kahweol causes an increase in total cholesterol as well as low density lipoprotein (LDL) cholesterol, triglycerides, and alanine aminotransferase (ALT) activity.” It also explains why a filter is the lever at all: “Brewing releases oil droplets containing the two compounds from the ground coffee beans” (NIEHS/NTP, Cafestol and Kahweol: Review of Toxicological Literature, 1999, PDF).
The same review summarises a controlled human trial: volunteers drinking 0.9 L of cafetière (French press) coffee a day for 24 weeks, delivering 38 mg of cafestol and 33 mg of kahweol, showed LDL cholesterol raised by 9 to 14% above baseline relative to a filtered-coffee group whose brew provided under 1 mg of the diterpenes. Two details deserve equal billing: ALT activity rose too, and “All increases, however, were reversible upon withdrawal of treatment.”
The amounts, measured recently. A 2025 study analysed coffees by liquid chromatography–mass spectrometry and reported median cafestol and kahweol in mg/L as 176/142 for workplace brewing machines, about 90/70 for percolator and French press, 28/21 for boiled coffee through a fabric filter, and 12/8 for home-brewed paper-filtered coffee (Orrje et al., Nutrition, Metabolism and Cardiovascular Diseases, 2025). That is roughly an order of magnitude between paper and a French press.
Three honest caveats, because this is a health claim and it should not be oversold:
- The evidence is about LDL cholesterol. It is not a demonstration that switching filters prevents a heart attack.
- The trial dose was 0.9 litres of French press coffee a day for six months. One cup a day is a much smaller exposure, and the effect reversed when the coffee stopped.
- The two sources do not agree on everything. The 1999 review calls percolated coffee “negligible” in diterpenes; the 2025 measurements put a percolator near a French press. Where they disagree we have used the newer measurements.
One footnote that surprises people: decaf does not help here. The NIEHS review notes that “Regular and decaffeinated coffees also have similar diterpene contents.” This is an oil question, not a caffeine question.
Paper, metal and cloth on flavour
The mechanism is the same one as above: the more oil and fine sediment a filter holds back, the cleaner and thinner the cup, and the more it lets through, the heavier and muddier. The National Coffee Association orders the three materials this way: “Cloth filters tend to absorb more oils and sediments than a metal filter but less than paper filters”, and “Generally, paper filters absorb more of the oils in coffee, which can contribute to a cleaner, lighter taste” (NCA, Cold brew coffee).
An earlier version of this page had cloth in the wrong place. It said cloth filters “allow some oils to pass through” but “filter out most of the fine particles, resulting in a clean and smooth brew” — which put cloth closer to metal on oils. Both the trade body’s ordering and the 2025 diterpene measurements put cloth much closer to paper. That table row has been corrected.
| Paper | Cloth | Metal | |
|---|---|---|---|
| Oils and sediment retained | Most | Between, nearer paper | Least |
| Cup character | Clean, light, transparent | Smooth, some body | Heavy, full, some silt |
| Cafestol and kahweol in the cup | Lowest measured (12/8 mg/L) | Low (28/21 mg/L through fabric) | High (French press about 90/70 mg/L) |
| Reuse | Single use | Reusable, needs careful cleaning | Reusable, effectively indefinitely |
| Main failure mode | Running out of them | Souring or mould if stored damp | Oil build-up dulling the taste |
None of the flavour descriptions above are our tasting notes. We have not run a blind comparison, and any page that gives you one without describing how it was run is inventing it.
Bleached versus unbleached paper
White filters have been whitened; brown ones have not. That is the difference. We found no published measurement of a taste difference or a safety difference between the two in brewed coffee, and rather than fill that gap with an opinion we will say so.
The worry people have usually heard is about dioxin. It is worth knowing where that concern actually lives: dioxin (2,3,7,8-TCDD) is a regulated priority pollutant in the effluent of pulp and paper mills, controlled by the EPA under its Pulp, Paper and Paperboard Effluent Guidelines, amended by the 1998 Cluster Rule. That is a rule about what mills may discharge into water, not a measurement of what is in your filter or your cup, and we are not going to present it as one. If you would rather avoid chlorine bleaching entirely, oxygen-bleached and unbleached filters are both sold everywhere.
Rinsing the filter
The old version of this page said to always rinse a paper filter “to remove any paper taste”. Rinsing is a reasonable habit, but be clear about which parts of that are established. Preheating is the part that definitely does something: pouring hot water through the filter and the brewer warms the cone or the carafe so your brew water is not immediately cooled by cold ceramic or glass, and it seats the paper against the wall so water does not run down the gap. The paper-taste part we could not find a measurement for. Some people taste it, some do not, and there is no published blind test we can point you at.
Cloth is the one that genuinely needs a routine, because it holds coffee oil and stays damp. The NCA’s instruction is specific: “Rinse and dry cloth filters thoroughly; you can use mild detergent to wash if preferred. Every six to eight weeks, boil the filters in fresh water for 10 minutes.” Metal filters need oil removed periodically or the residue turns rancid and shows up in the cup. Paper needs nothing, which is most of its appeal.
Filter sizes, and what the numbers mean
The old FAQ on this page confidently mapped #1, #2 and #4 cone filters to specific numbers of cups. Those numbers were unsourced, and they run into the same ambiguity that causes most bad coffee advice: a brewer manufacturer’s “cup” is not a fixed volume, and in coffee a cup is conventionally 6 fl oz rather than 8. The reliable answer is dull — the size is stamped on the box, and your brewer’s manual names the size it takes. Match those two and ignore the cup counts.
So which filter should you use
- Watching your LDL, or told to? Paper, and preferably not a daily French press. This is the only part of the choice with a measured health consequence.
- Want body and texture? Metal. You are drinking the oils on purpose; that is the point of a French press.
- Want to stop buying filters without losing much clarity? Cloth, if you will actually keep it clean. If it will live damp in a drawer, buy paper.
- Brewing cold? The NCA notes that “Cloth is the most common material for cold brew coffee filters, though some brewers do support the use of paper filters for additional filtration.” Our page on cold brew versus nitro coffee covers the rest of that method, including the concentrate question.
Whatever you choose, filter type is a smaller lever on taste than dose, grind and water temperature. If your coffee is not right, start with those: our house ratio is 55 g of coffee per kilogram of water (1:18 by weight), the roast question is covered in coffee roast levels, and the sourced brewing temperature range lives on best temperature for French press coffee.
Disclaimer: This page is general information, not medical advice. The cholesterol findings described here come from published research on groups of people and do not predict what will happen to any individual. If you are managing high cholesterol, or taking medication for it, decide with your own clinician rather than on the basis of a coffee article.
