Coffee and Cholesterol: How Much Your Brew Method Actually Matters

7 min read · Updated August 9, 2026

The AHA named unfiltered coffee as an LDL risk. Here is how much cafestol each brew method delivers, what the trials measured, and when the difference is worth acting on.

The July 2026 American Heart Association scientific statement on caffeine did something unusual for a cardiology document. It named brewing equipment.

Cafestol, it says, sits in “unfiltered coffee, including French press, Greek/Turkish, and Scandinavian boiled coffee, and to a lesser extent in espresso-based and Moka coffee,” and it is “not present in substantial amounts in paper-filtered or instant coffee” (Marcus et al., Circulation 2026). Of everything in that statement, this is the part a home brewer can act on in the morning, and it is the one piece backed by randomized trials rather than population surveys.

This guide is educational and not medical advice. The numbers below are worth knowing, and how they apply to you is a conversation with your clinician.

What cafestol is

Cafestol and kahweol are diterpenes, oily compounds that make up roughly 1 percent of the weight of a coffee bean. Hot water pulls them out of the grounds into oil droplets and suspended fines. Cafestol is the one that matters for lipids, and it has been described as the most potent cholesterol-raising compound in the human diet (Urgert and Katan, 1997).

It works by suppressing LDL receptor activity and interfering with bile acid synthesis, so less LDL gets cleared from the blood. Caffeine has nothing to do with it. Decaf brewed in a French press carries cafestol just the same, which is why the AHA describes it as a component of both caffeinated and decaffeinated coffee.

How much cafestol each method delivers

The cleanest cross-method comparison standardizes everything to a 120 mL cup (Naidoo et al., 2011).

Paper-filtered drip sits about 35 times below Turkish coffee. That gap is the whole story, and everything else is detail.

A caution on units. Published measurements vary several-fold between studies because cup volumes, roast levels, grind, and contact time all move the number. A Swedish study using different units found paper-filtered brews at 12 mg/L against roughly 90 mg/L for French press and 939 mg/L for boiled coffee (Orrje et al., 2025). The absolute values disagree with the table above. The ranking does not.

What a paper filter is actually doing

Most people picture the paper as a screen catching oils. The measurements tell a more interesting story. When researchers tracked where the cafestol in the grounds ended up, the spent coffee bed retained 87.45 percent, the paper filter caught 12.41 percent, and 0.15 percent made it into the cup (Rendón et al., 2017).

The bed of grounds does most of the work. The paper handles the last stretch.

That explains a result which otherwise looks odd. Metal and cloth filters measure far closer to paper than to French press, because they still brew through a grounds bed. In the same drip geometry, a nylon mesh filter passed 0.64 mg per cup against 0.08 mg for paper, an eight-fold difference that still leaves metal-filtered drip well under a cafetière (Naidoo et al., 2011). Paper filter type matters too. Micro-perforated papers and finer grinds push the residue up, and paper-filtered brews from six countries ranged from 1.62 to 2.98 mg/L (Rendón et al., 2018).

What the trials measured

Cafestol raises cholesterol in a straight line with dose. Pooling 11 supplement trials gives about 0.13 mmol/L (5 mg/dL) of serum cholesterol per 10 mg of cafestol per day, with roughly 80 percent of that rise landing in LDL (Urgert and Katan, 1997).

Now put real coffee through that. The best real-brew experiment randomized 46 adults to five or six strong cups a day, cafetière against filtered, for 24 weeks. LDL ran 9 to 14 percent higher in the cafetière group, still 0.26 mmol/L higher at 24 weeks, about 10 mg/dL (Urgert et al., BMJ 1996). Liver enzymes rose too, with ALT above the upper limit of normal in 8 of 22 cafetière drinkers. Every change reversed after stopping, and the full effect took 8 to 12 weeks to develop.

Then there is the trial that complicates the picture. A 2024 randomized, placebo-controlled study gave 40 people 12 mg of cafestol a day for 12 weeks, a dose its authors put at about four 150 mL cups of French press. LDL did not move: a between-group difference of −0.06 mmol/L, 95 percent CI −0.30 to 0.18 (Mellbye et al., Nutrients 2024).

Read those two together before drawing a conclusion. The dose-response equation predicts about 0.16 mmol/L from 12 mg a day, which sits inside that trial’s confidence interval. A study of 40 people was never going to separate an effect that small from zero. Heavy unfiltered drinking has a demonstrated effect. A moderate habit sits in a range where the effect is small enough that a 40-person trial cannot resolve it.

Where espresso actually lands

The AHA statement groups espresso with unfiltered coffee, which reads alarmingly if you pull two shots a day. Two things pull in opposite directions.

Concentration is high. Serving size is small. Scaling the 1.2 mg per 120 mL figure down to a 40 mL double shot gives about 0.4 mg of cafestol. A separate measurement is less flattering, putting total diterpenes in espresso at 30.7 to 58.8 mg/L, or 1.2 to 2.3 mg per 40 mL shot (Moeenfard et al., 2015). Those two disagree by several-fold, which is normal in this literature. Take the higher one and a double shot still sits below a single 240 mL mug of French press, which carries about 5.6 mg. The AHA’s own wording lands in the same place, putting espresso and moka beneath the unfiltered methods.

Population data still finds a signal. The Tromsø Study looked at 21,083 Norwegians and found 3 to 5 espressos a day associated with total cholesterol higher by 0.16 mmol/L in men and 0.09 mmol/L in women, against 0.25 and 0.18 for the same intake of boiled or plunger coffee (Svatun et al., Open Heart 2022). That study is cross-sectional and measured total cholesterol rather than LDL, and its authors say plainly that the design limits causal inference.

One more data point for the daily-driver question. A Norwegian cohort of 508,747 people followed about 20 years found filtered coffee associated with lower total mortality than no coffee, HR 0.85 in both sexes, while unfiltered coffee showed a weaker association (Tverdal et al., 2020).

What to do with this

For most people the practical answer is small.

• If your LDL or ApoB is fine, brew what tastes good. One or two unfiltered cups a day is a fraction of the dose that moved lipids in trials.
• If your LDL or ApoB runs high, make paper-filtered coffee the everyday cup and keep unfiltered methods for when you want them. This is a change you can make this week, and unlike most diet advice it has randomized evidence behind it.
• If you drink four or more unfiltered cups a day, this is the group the trials were actually about. Switching the bulk of that volume to paper is the highest-value change on this page.
• Give it a season. The effect takes 8 to 12 weeks to build and the same to unwind, so a lipid panel two weeks after switching filters tells you very little.
• Decaf changes nothing here. Method drives the cafestol dose.

Volume matters more than method for most drinkers, because the dose scales with how much coffee you actually drink. A single French press cup and a 20-ounce unfiltered mug are different exposures.

Track it in BeanBench

The variable that matters here is one you already log. In BeanBench you can record brew method and filter type on every brew, so if you change your daily driver you can point at the date you switched and see how long you have held it. Pair that with the wider health picture in is coffee good for you, compare the two families of method in pour over vs. espresso, or pick a paper-filtered V60 recipe to make the everyday cup.