日本語版

Topic hub — 2026-09-07

Coffee bitterness

A considerable part of coffee's bitterness is made by the process of roasting. Chlorogenic acid, abundant in green beans, lactonizes under the heat of roasting into quinides, and if roasted still more strongly, part of that goes on to decompose into a group of compounds with a phenylindane skeleton. At the same time, green beans already contain bitter substances such as caffeine, trigonelline, and mozambioside, and roasting both destroys and remakes these. An account that explains everything with caffeine alone collapses these two layers into one. In fact, one report found that an extract of green beans showed no significant bitterness even with caffeine present at more than 10 times its detection threshold, and that adding caffeine to roasted coffee did not move the bitterness sensory score. Cultivar and the mixing-in of defect beans, brewing conditions, and the addition of sugar or milk each move bitterness in their own separate place. Here, we line up, as far as it could be confirmed, the claims about what roasting makes and what moves bitterness elsewhere.

When a brewed cup tastes bitter, separate adding something to this cup from changing the next brew — that makes the choice easier. Brewing and what you add reads a comparison that added milk or oat drink and a comparison that changed the brew water temperature in the same condition table.

What this page establishes

• The pathway by which the heat of roasting turns chlorogenic acid into quinides, and deeper roasting turns those into harsh-bitter compounds, has been confirmed by multiple independent studies.
• For the share of a cup's bitterness that caffeine accounts for, a secondhand-cited figure puts it at up to 10-30%, but in one study that actually added caffeine to roasted coffee, the bitterness sensory score did not increase significantly.
• Which of robusta and arabica tastes more bitter has not yet settled on one answer: the ranking flips between a comparison with matched roast conditions and a comparison optimized bean by bean.
• Changing only the brew water temperature did not move bitterness significantly in one study that held concentration fixed. In a separate study that varied water temperature widely without fixing concentration, bitterness increased significantly.

Bitterness made by roasting

Chlorogenic acid, abundant in green beans, lactonizes under the heat of roasting into quinides. In roasted coffee, 7 kinds of quinides have been identified, and the amount formed has been reported to reach up to about 30% of the precursor.

Roasting still further, part of the quinides decompose into a group of compounds with a phenylindane skeleton. In a model system where caffeic acid was heat-treated, 8 compounds in this series were identified, with recognition thresholds ranging from 23 to 178 µmol/L.

The correspondence between roast degree and bitterness has also been confirmed by sensory scores from trained assessors. As roasting deepened across Agtron color values 73, 62, and 53, the bitterness score rose significantly.

Which aroma stands alongside the bitterness also changes with roast degree. Tracking which sensation comes to the fore over the course of drinking, the partner to bitterness switches between medium and dark roast: roasty aroma for the former, burnt aroma for the latter.

Bitter substances already in the green bean

Green beans, too, already contain substances involved in bitterness. Trigonelline, an alkaloid, was detected at an average of 8.3 mg/g in direct measurement of 277 single green beans.

Mozambioside, a glycoside almost entirely restricted to arabica, is also one of the bitter substances in green beans. It is detected in the range of 396 to 1188 nmol/g in arabica green beans, while in robusta it stays below 5 nmol/g.

And yet, even with all these bitter substances present, the green bean extract itself is not necessarily bitter. Adding 7.5 mmol/L of caffeine to an unroasted extract made from Colombian green beans — a concentration more than 10 times the detection threshold of 0.67 mmol/L — did not produce any significant detected bitterness.

How much of a cup's bitterness is caffeine?

For how much of a cup of coffee's bitterness caffeine accounts for, a secondhand-cited figure is often quoted. Citing a 1979 study, a figure is given stating that caffeine already present in green beans contributes at most about 10 to 30% to the overall bitterness of a coffee beverage. This figure is not from the original work itself but a statement made by a later paper citing that study — it is a secondhand citation.

In an experiment that actually checked this on roasted coffee, this contribution looks even smaller. Even when 7.5 mmol/L of caffeine was added to decaffeinated roasted coffee, the bitterness sensory intensity moved from 3.8 to 4.0 — not a significant difference.

The panel could identify the addition of caffeine as bitterness only when it exceeded the concentration coffee can naturally reach. In a three-alternative forced-choice test, it could not be detected at 7.5 or 10.0 mmol/L, and a significant increase in bitterness intensity was detected only at 12.5 mmol/L.

There are also components that weaken caffeine's bitterness. When the high-molecular-weight fraction of coffee (including melanoidins) was combined with caffeine, the bitterness sensory intensity dropped to about half, to 2.5, from a reference of 5.0 for caffeine alone.

Cultivar and defect beans — robusta, arabica, quaker beans

Differences in bitterness by cultivar change depending on how the roast conditions are matched. Comparing robusta and arabica espresso roasted with the same roasting process, robusta scored higher on bitterness and aroma persistence.

However, there is an example in which the order flips in a comparison where roast conditions were individually optimized for each bean. A decaffeinated arabica blend was rated more bitter than a robusta blend, while robusta had stronger astringency. Because the two reports do not stand on the same premise (the same roast degree), we lay out both rather than favoring one.

In terms of composition, robusta tends to have higher caffeine and chlorogenic acid content than arabica. Among 6 roasted bean samples compared, Brazilian robusta had the highest caffeine content of the group, at 22.38 mg/g.

Not only cultivar but also the mixing-in of defect beans moves bitterness. In a sensory evaluation that mixed quaker beans (beans from immature seeds) into natural-processed specialty coffee, quaker beans with an Agtron color of 82.8 or above significantly increased bitterness and astringency once contamination reached 7 beans (10.8%) or more out of 65 in a single cup (a report within the scope of the abstract).

Brewing and what you add

What to change when coffee is bitter

For a cup already brewed, the comparison that added milk or oat drink is the closer match. For the next cup, the comparison that changed the brew water temperature is the closer match. That said, the answer for water temperature changes with the strength the study held fixed, so it cannot be decided that a lower temperature is always less bitter.

Adding milk, adding sugar, and changing water temperature are each separate comparisons. For a cup that is a problem right now, read the addition rows; for the next brew, read the temperature row. There is no household-common amount to add or brewing procedure here.

To check the difference in this cup, before changing the whole amount, a small portion can be split into a separate cup and compared with milk or oat drink added against the cup left as is. This is not a procedure that makes the study's blend a household correct answer — it is an editorial suggestion for checking the balance of taste.

Under conditions where strength was matched, varying the brew water temperature across 87, 90, and 93°C produced no statistical difference in the trained panel's bitterness score.

The result differs in a comparison that does not fix strength. Brewing dark-roast coffee at 4, 22, and 92°C, bitterness increased significantly at 92°C, and total dissolved solids rose as well. Sourness was also higher at the same 92°C, moving in the same direction as bitterness.

In a study that added sucrose to coffee, a phenomenon in which taste and flavor suppress each other was recorded. However, from the abstract alone that could be confirmed, the number of participants, concentration, and the bitterness-only amount of change cannot be decided.

In a comparison that blended medium- and dark-roast coffee at 70% with milk or oat drink at 30%, the rate at which bitterness was chosen as the most salient sensation dropped across several time windows. This is not bitterness intensity itself but a measurement that selects the most salient sensation at each point in time.

How bitterness gets talked about

How bitterness itself is talked about is also split by industry and product. The official SCA cupping handbook records that while there is a culture that softens strongly bitter, dark-roast, high-robusta coffee with sugar or milk, there is another culture that takes a different approach, lowering the bitterness level itself by using light- or medium-roast arabica.

Expectations of specialty coffee are not monolithic either. In a survey of 1,005 Brazilian consumers, bitterness (41.6%) was the 2nd most cited basic taste expected in specialty coffee, after sweetness (71.4%). The researchers add that this type of coffee has lower bitterness than commodity coffee.

Even on the canned-coffee shelf, how bitterness is talked about can split within the same brand. An unsweetened black line for mornings explicitly states its bitterness as "a crisp, bitter taste," while the description of the same brand's lightly sweetened line makes no mention of bitterness at all.

Frequently asked

Is coffee's bitterness the taste of caffeine?

Caffeine's contribution to a cup's bitterness has been measured from several angles, but none of them add up to a result that lets us flatly say "it's the taste of caffeine." In a three-alternative forced-choice test, the panel could not detect changes in caffeine within the 7.5-10.0 mmol/L range that coffee can naturally reach, and a significant difference appeared only once it was raised to 12.5 mmol/L, a level that does not occur naturally. In a separate experiment that added caffeine at 10 times the threshold to an extract made from unroasted green beans, bitterness likewise did not appear. The widely cited figure of up to 10-30% for the contribution rate, known to be a secondhand citation, is not from the original work itself but a value cited by a later paper — it is not a direct experimental value.

Why does darker roasting taste more bitter?

The explanation is progressing on the side of roasting's chemical changes. Chlorogenic acid, abundant in green beans, lactonizes under the heat of roasting into bitter substances called quinides. Up to this point the reaction happens preferentially at light-to-medium roast; 7 kinds of quinides have been identified, and the amount formed has been reported to reach up to about 30% of the precursor. Roasting still more strongly, part of these quinides decompose into a different group of bitter compounds. Sensory scores from trained assessors have also confirmed that the bitterness score rises significantly as roasting deepens. That said, for the size of each component's contribution, many reports remain confirmed only at the abstract level.

Robusta or arabica — which is more bitter?

The answer changes depending on how the roast conditions are matched — that is the actual state of measurement so far. In a study comparing robusta and arabica espresso roasted with the same industrial roasting process, robusta was rated higher on bitterness sensory intensity and aroma persistence. Yet in a separate study that individually optimized roast conditions for each bean, a decaffeinated arabica blend was rated more bitter than robusta, while robusta had stronger astringency. On the composition side, among 6 roasted bean samples compared, Brazilian robusta had the highest caffeine content, at 22.38 mg/g, but this is a separate measurement from the sensory evaluation of bitterness itself.

If you have brewed coffee too bitter, does adding milk or sugar change it?

In a comparison of coffee with milk or oat drink added, the rate at which bitterness was chosen as the most salient sensation dropped. There are reports pointing toward a decrease, but attention is needed to how differently they measure it. In an experiment adding milk or oat drink to medium- and dark-roast coffee, the rate at which bitterness was chosen as the dominant sensation dropped significantly across several time windows. However, this comes from a method that picks the most salient sensation at each moment in time, not a direct measurement of bitterness strength itself. In a separate study that added sucrose, hypoadditivity has been reported in which overall taste/flavor intensity is suppressed by roughly 30-40%, but this remains confirmed only at the abstract level. Neither measurement method selectively erases bitterness.

What is not known

• We have not been able to reach an independent study that directly measured the exact share of a cup's bitterness that caffeine accounts for. The secondhand-cited 10-30% remains with its original source unconfirmed.
• The mechanism by which coffee's high-molecular-weight fraction suppresses caffeine's bitterness has only been confirmed for a single fraction, and generalization to other roast degrees and brewing methods has not yet been measured.
• No study has been found that directly compared the bitterness ranking of robusta and arabica under the same roast degree and the same brewing conditions.
• The value treated as the threshold at which mixing in immature beans increases bitterness comes from a single study, and reproduction has not yet been measured.

Read in the encyclopedia

The encyclopedia entries below are in Japanese.

• coffee brew — a drink whose bitterness is reshaped not just by caffeine but by roasting byproducts and brewing conditions
• caffeine — a bitter substance present in both green and roasted beans; its standalone contribution can fail to register significantly before or after roasting
• trigonelline — a pyridine alkaloid abundant in green beans; most of it decomposes during roasting
• Mozambioside — a bitter furokaurane glycoside specific to arabica green beans; roasting produces breakdown products from it
• chlorogenic acid — a phenolic compound abundant in green beans; it decreases with roasting and gives rise to breakdown products including quinides
• melanoidins — a high-molecular-weight fraction formed by roasting's browning reactions; it works to weaken caffeine's bitterness
• green coffee bean — the unroasted bean, containing trigonelline, caffeine, and chlorogenic acid
• robusta coffee / arabica coffee — the two species whose bitterness ranking changes depending on how roast conditions are matched
• coffee roasting — the process that makes bitterness and remakes the underlying compounds

Go deeper in the research notes

• Coffee's Bitterness Was Never Just Caffeine — follows the leading players in roasting-made bitterness: quinides, 4-vinylcatechol oligomers, mozambioside breakdown products, and more
• Roasting Makes Bitterness, and Dark Roast Changes It — follows the mismatch in roast-degree naming, the report that bitterness scores rise with darker roast, and the compounds that swap places behind it
• Butter Roasting and Southeast Asian Kopi — Roasting That Wraps Up Bitterness — a record of a roasting method that adds fat and sweetness to robusta beans, laid alongside the ways bitterness gets talked about

Source ledger

The primary information and limitations behind the claims listed here can be checked from the source link inside each card. The full list is gathered in the Source ledger.