日本語版

Topic hub — 2026-10-21

Does roasting tea (stronger firing) reduce its bitterness?

A caveat first: we found no study, within our search, that had people compare bitterness alone across roasting strengths for Japanese hojicha or sencha firing. One sencha study rated "bitter-astringent taste" (bitterness and astringency together) and found no consistent trend. The full-text studies that rated bitterness as a separate attribute and are included in the comparison are on oolong tea (Tieguanyin and Wuyi rock tea): one detected no significant difference and others report an increase. The point of this page lies in what has not been established. What do studies that changed roasting strength and had people compare tea bitterness say? This page lines up the included studies together with their conditions: type of tea, roasting temperature and time, brewing, and who rated it on what scale. It then separates what was established within one experiment, what differences might explain, and what this comparison alone cannot separate. It is about perceived taste, not advice on choosing or brewing tea or on health. Differences in brewing, such as water temperature or cold brewing, are covered on the green tea temperature page.

What this page establishes

• That no study rating bitterness alone was found for Japanese tea, and a study rating bitter-astringent taste across sencha firing strengths found no consistent trend; other Japanese sources rated aroma, colour, overall scores or preference.
• That in oolong, extending roasting time produced no detected significant difference in bitterness, while a study comparing unroasted and deep-roasted tea found a higher score for the deep roast (which pair differs cannot be identified).
• That even a study whose abstract says roasting weakened bitterness detected no significant difference in bitterness in its text and figure.
• That bitterness did not necessarily weaken even when catechins, regarded as bitter compounds, fell with roasting, and the compounds that strengthened bitterness have not been confirmed.

What the included studies allow us to say

• We found no study, within our search, in which people rated bitterness alone across roasting or firing strengths for Japanese tea. 1 study (Nakagawa et al. 1979, 10-odd tea station staff and about 30 students) rated "bitter-astringent taste" (bitterness and astringency together) for sencha fired strong, medium and weak; some sets did not line up in that order, and the authors report no consistent trend within the firing range they set by expert judgement. Other Japanese sources rated aroma, colour, overall scores or preference. [rows: Nakagawa 1979 Strong vs weak firing (high-grade sencha, students); Nakagawa 1979 Strong vs weak firing (high-grade sencha, tea station staff); Nakagawa 1979 Strong vs weak firing (low-grade sencha, students); Nakagawa 1979 Strong vs weak firing (low-grade sencha, tea station staff)]
• Among full-text oolong studies in which people rated bitterness across roasting strengths, one detected no significant difference and others report an increase. Cao et al. 2021 detected no significant difference in bitterness when roasting at 105 °C was extended from 1 h to 5 h (about 2.7 and about 2.3–2.4, read from the figure). Gao et al. 2022 found a higher score for deep-roasted than unroasted tea (about 2.3–2.4 and about 3.0, 7 assessors); the bitterness axis carries a significance mark, but which pair differs cannot be identified. A Wuyi rock tea study writes that bitterness rose when roasted up to the 4th time. [rows: Cao 2021 105 °C for 5 h vs 1 h; Gao 2022 Deep roast (2 roastings) vs unroasted; Feng 2026 Up to the 4th roasting vs up to the 3rd]
• These studies support neither "roasted tea is not bitter" nor "roasting makes tea bitter". Cao et al. 2021, whose abstract says roasting weakened bitterness, detected no significant difference in bitterness in the text and figure. One study states in its full text that bitterness fell (Lu’an Guapian green tea), but it has no separate bitterness score or test, so it is not included in the comparison. Several other studies report reduced bitterness in the saved summaries of their abstracts, but neither the original abstracts nor the full texts have been checked. [rows: Cao 2021 105 °C for 5 h vs 1 h]

Bitterness ratings at a glance

Human bitterness ratings only. All values are read by eye from figures in the papers, not taken from tables. Even where the score range is the same, the scale descriptions, assessors, brewing and roasting differ, so values are not compared across studies. "Direction" shows which was more bitter; for rows with no values used, it follows the text and the figure. Bitter-astringent rows are not ratings of bitterness alone and are in the detailed tables; whether the difference is significant is shown by the test in the same "Direction / test" column. Select a study name to jump to its detailed table. Rows on astringency and composition are in the detailed tables by study.

Results in which people rated bitterness only. Where each value comes from (a table or a reading from a figure) is shown in its row; scales and methods differ between studies, so compare within a row, not across rows.

Conditions and results, study by study

For each study, the conditions for the tea, roasting, brewing and who rated it, and the results. Each row is one study × one measured result, so different comparisons in the same study are separate rows. Under each row we give where it is in the paper and whether it can be used to judge whether roasting the leaves or stronger firing changes the bitterness people perceive in tea, with the reason. Rows on astringency or composition alone are kept so that it is clear they cannot be used for bitterness. Nothing is left blank: "not given", "not read" and "not applicable" are kept apart.

What was established within an experiment

Only results observed in comparisons within one study. Human ratings and compound measurements are kept apart.

• When only the final roasting time varied for Tieguanyin from the same leaves and process, astringency fell significantly (about 3.0 to about 2.3), but no significant difference in bitterness was detected (about 2.7 and about 2.3–2.4). Adding 1 h at 130 °C after 5 h also produced no detected significant difference in bitterness. [rows: Cao 2021 105 °C for 5 h vs 1 h; Cao 2021 105 °C for 5 h vs 1 h (astringency); Cao 2021 5 h plus 1 h at 130 °C vs 5 h]
• When the same Tieguanyin was deep-roasted in 2 stages, the bitterness score rose (about 2.3–2.4 to about 3.0; the bitterness axis carries a significance mark, but which pair differs cannot be identified), and total monomeric catechins fell (98.09 to 80.74 mg/g, letters a and c in the table). No significant difference in caffeine was found between unroasted and deep roast (16.73 and 16.54). [rows: Gao 2022 Deep roast (2 roastings) vs unroasted; Gao 2022 Deep roast vs unroasted (total monomeric catechins); Gao 2022 Deep roast vs unroasted (caffeine)]
• When only firing degree varied for sencha of the same grade, the bitter-astringent score for strong versus weak firing was higher or about the same for strong in every set (students, high grade 4.06 and 3.56; staff, low grade 3.30 and 2.78, among others), but including medium, some sets were not in order (students, high grade: medium lowest at 3.33). For staff and high-grade sencha the strong-weak difference was 0.01. No test is reported. [rows: Nakagawa 1979 Strong vs weak firing (high-grade sencha, students); Nakagawa 1979 Strong vs weak firing (high-grade sencha, tea station staff); Nakagawa 1979 Strong vs weak firing (low-grade sencha, students); Nakagawa 1979 Strong vs weak firing (low-grade sencha, tea station staff)]

Candidate explanations for the differences

Conditions that differ between studies and could explain the differences in bitterness results. Several conditions differ at once, so which one mattered is unknown. These are candidates, not causes.

• Because the bitterness score did not fall even though catechins decreased, Gao et al. suggest that Maillard products formed during roasting (such as 5-HMF) may have strengthened bitterness. The compounds that strengthened bitterness have not been confirmed. [rows: Gao 2022 Deep roast (2 roastings) vs unroasted; Gao 2022 Deep roast vs unroasted (total monomeric catechins)]
• The difference between the 2 studies may also reflect how roasting was varied. Cao et al. kept 105 °C and extended time from 1 h to 5 h; Gao et al. compared unroasted tea with tea roasted at 107 °C after 12 days and again at 115 °C after a further 22 days. The starting teas also differed, and the panels had 9 and 7 members; whether assessors overlapped is unknown. [rows: Cao 2021 105 °C for 5 h vs 1 h; Gao 2022 Deep roast (2 roastings) vs unroasted]

What is not known — what this comparison alone cannot separate

• Japanese hojicha roasting (high temperature for a few minutes in a roaster) and oolong roasting (lower temperature for hours to tens of hours) differ in temperature, time and starting tea. Whether the oolong results apply to hojicha or to roasting sencha at home is unknown. [rows: Cao 2021 105 °C for 5 h vs 1 h; Gao 2022 Deep roast (2 roastings) vs unroasted]
• The bitterness scores used are read by eye from the figures of Cao et al. and Gao et al.; compound values are copied from tables. For Gao et al. which pairs differ cannot be identified, and for the Wuyi rock tea study no values are used (only the direction from the text and figure). The panels had 9, 7 and 7 members, and the bitter-astringent score of Nakagawa et al. does not separate bitterness. For none of the studies can repetition across production lots be confirmed adequately from the description. [rows: Cao 2021 105 °C for 5 h vs 1 h; Gao 2022 Deep roast (2 roastings) vs unroasted; Feng 2026 Up to the 4th roasting vs up to the 3rd; Nakagawa 1979 Strong vs weak firing (high-grade sencha, students)]

How studies and rows were counted

The number of evidence rows is not the number of studies. Table rows, ledger evidence rows, papers and research groups are counted separately.

• Sources in the table: 4, of which 3 have results on people's bitterness ratings in this table; for the others, only results on something else (compounds, smell) are shown, or they are used only for context.
• Rows in the table (study × measured result): 12. Rows on perceived bitterness: 4.
• Evidence rows in the lab's ledger these rows rest on: 1. One paper can supply several rows. Rows without a ledger evidence row come from other tables of the same papers, or from sources newly read for this question.
• Research groups among the studies with bitterness results in this table, as grouped in the ledger: 2 (reports from the same lab count as 1; co-authors and affiliations can overlap, so this is not a count of independent replications).

What we looked at

The sources examined for this question and why each was included or excluded. We searched J-STAGE and Europe PMC for studies rating bitterness across roasting or firing strengths of tea (hojicha, bancha, sencha, oolong and others), and added the sources already in the lab's ledger (CiNii and conference abstracts were not searched). Results documents from a prefectural tea experiment station are not treated as peer-reviewed papers, and the type of each source is stated. Some studies report reduced bitterness in the saved summaries of their abstracts, but neither the original abstracts nor the full texts have been checked. The search described here is a record, not a claim that nothing was missed.

Read in the encyclopedia

• Hojicha — roasted Japanese tea
• Tieguanyin — the tea in the included studies
• Tea firing (hiire) — heating that dries the leaves and adds aroma
• Tea's bitterness and astringency — the tea bitterness hub
• Is green tea less bitter brewed cooler or cold? — a comparison of brewing

Source ledger

Each study's source can be reached from its section above. All of the lab's sources are listed in the source ledger.