Matcha vs Coffee: The Theanine Trials Didn't Use Matcha

Matcha went from tea ceremony to drive-through in about a decade, and it arrived carrying a specific promise: the alertness of coffee without the edge. Clean energy, clean focus, no crash. That is a marketing phrase, but it is not pure invention — there is a real compound behind it, and there are controlled trials behind the compound. What is shakier is the distance between those trials and the cup in your hand.

What The Composition Review Actually Says

Matcha is a powdered green tea, and the paper most often cited for what is in it, Kochman 2020, is a narrative literature review rather than a measurement study. Every composition figure in it is quoted from another author's work and cited individually. That distinction matters, because those quoted figures end up doing most of the work in the matcha-versus-coffee argument.

The review does describe the agricultural step matcha producers are known for: for the majority of the growth period the bushes are covered with bamboo mats to shade the leaves from excessive direct sunlight, and shaded plants are reported to produce higher amounts of amino acids and bioactive compounds, including chlorophyll and theanine Kochman 2020. Grade, cultivar and harvest timing plausibly contribute to the variation between one tin and the next as well, but shading is the driver this review actually names.

One claim that circulates in matcha's favour is worth correcting, because it is usually attributed to this review. The review's only comparison between steeped and powdered tea concerns polyphenols: it reports that infusions made by steeping tea leaves have a lower polyphenol content than those made from the powdered form, and suggests the grinding process itself may accelerate the extraction of polyphenolic compounds Kochman 2020. It does not extend that comparison to caffeine or to amino acids, and it describes matcha throughout as consumed as an infusion of the powder. The familiar “you consume the whole leaf, so you get more of everything” framing is not a statement this source makes. The same authors write that the direct impact and mechanisms responsible for matcha's properties have not been sufficiently explored, and they call for detailed compositional analysis, cell and animal work, and randomised clinical trials before recommendations for increased matcha consumption are made Kochman 2020.

The Caffeine Arithmetic

The most common claim about matcha is that it simply carries less caffeine. Sometimes it does. But the honest framing is arithmetic rather than chemistry: your dose is the mass of powder or ground bean used, multiplied by its caffeine concentration. Change either number and the comparison flips.

Concentration varies, and the ranges overlap more than the marketing suggests. Kochman 2020 quotes 18.9 to 44.4 mg of caffeine per gram for matcha and 11.3 to 24.67 mg/g for green teas generally, alongside 10.0 to 12.0 mg of caffeine per gram for most coffee beans. Matcha is the more concentrated tea powder in those figures, but the two tea ranges overlap between 18.9 and 24.67 mg/g — roughly the top two-fifths of the green-tea range and the bottom quarter of matcha's — and each range is quoted from a single cited source rather than pooled.

Firmer anchors come from trials that weighed and analysed the product they fed people. Each of those trials was run or co-authored by scientists at the Central Research Institute of ITO EN, Ltd., a Japanese green tea and matcha company whose own product category was under test — a tie worth weighing in exactly the way the tea-industry and theanine-producer affiliations disclosed further down are. In a 12-week randomised trial, a daily matcha dose of 2070 mg supplied 66.2 mg of caffeine, 48.0 mg of theanine and 170.0 mg of total catechins Baba 2021. In a 4-week randomised trial, 2.7 g of matcha daily contained 71.5 mg of caffeine, 50.3 mg of theanine and 301.4 mg of catechins Baba 2024. In a 12-month randomised trial, 2 g of matcha taken in capsules supplied 66.2 mg of caffeine, 48.1 mg of theanine and 170.8 mg of catechins Uchida 2024. Three measured servings in the 2 to 2.7 g range, then, landed in the high sixties to low seventies of milligrams of caffeine.

The coffee comparator people reach for is usually EFSA's, and it needs handling with care. EFSA's caffeine opinion assumed that one cup of coffee contained 150 mL and 90 mg of caffeine when the original publication did not report it — a back-calculation convention for reading other researchers' trials, not a reference serving, and not the basis of EFSA's population exposure estimates, which applied measured caffeine concentrations to a European food-consumption survey database EFSA 2015. The same opinion notes that concentrations in coffee beverages depend on the manufacturing process, the type of beans and the type of preparation EFSA 2015. So 90 mg is a rough scale marker for a cup, not a fact about yours.

The safety statements in that opinion are worth knowing whichever drink you prefer, and their wording matters as much as their numbers. EFSA concluded that single doses of caffeine up to 200 mg — about 3 mg per kilogram of body weight — from all sources do not raise safety concerns for the general adult population, and that intakes from all sources up to 400 mg per day, about 5.7 mg/kg bw, do not raise safety concerns for adults in the general population except pregnant women, for whom intakes up to 200 mg per day do not raise safety concerns for the fetus EFSA 2015. EFSA also notes that single doses of 100 mg, about 1.5 mg/kg bw, may increase sleep latency and reduce sleep duration in some adults, particularly when consumed close to bedtime EFSA 2015. Two features of that get lost in repetition: the figures are scaled to body weight, so they do not transfer unchanged to a lighter person, and EFSA is describing intakes that do not raise safety concerns rather than issuing a recommended limit.

Those are totals, and they are indifferent to which plant supplied them. If you are using caffeine deliberately — before training, for instance — the dose and its timing matter more than the source, which we have looked at separately in our piece on when caffeine actually helps performance.

What The Combination Trials Found

L-theanine is an amino acid characteristic of the tea plant. Kochman 2020 attributes matcha's non-bitter taste to its relatively high theanine content, and attributes the umami sensation to theanine in combination with caffeine rather than to theanine alone.

The focus half of the story is usually built on the two most-cited acute crossover trials, and their arm structure matters more than their headlines. They are not the whole acute literature — the pooled review discussed below included 50 randomised controlled trials of tea, L-theanine, or L-theanine plus caffeine Payne 2025 — but they are the two the popular claim leans on. Haskell 2008 was a randomised, placebo-controlled, double-blind, balanced crossover in healthy volunteers with four conditions in a single session: placebo, L-theanine 250 mg, caffeine 150 mg, and the combination, with salivary caffeine co-monitored. Caffeine alone produced faster digit vigilance reaction time, improved Rapid Visual Information Processing accuracy, and attenuated increases in self-reported mental fatigue. The combination reproduced those two performance gains and added faster simple reaction time, faster numeric working memory reaction time and improved sentence verification accuracy, with reduced “headache” and “tired” ratings and increased “alert” ratings Haskell 2008. L-theanine on its own did not help: it increased headache ratings and decreased correct serial seven subtractions.

That is a more favourable result for the combination than a sceptical reading would like, and it should be stated plainly: the increase in subjective alertness in this trial sits with the combination arm, not with caffeine alone. What the trial does not establish is synergy. Only one effect is reported as a statistically significant interaction between the two compounds — delayed word recognition reaction time — while the rest are main-effect improvements Haskell 2008. The authors' own summary is measured: beverages containing L-theanine and caffeine may have a different pharmacological profile to those containing caffeine alone. The published abstract states neither how the compounds were delivered nor the sample size.

Owen 2008 compared 50 mg of caffeine, with and without 100 mg of L-theanine, against placebo in 27 participants, with each treatment separated by a 7-day washout and performance measured at baseline and again at 60 and 90 minutes. Caffeine improved subjective alertness at 60 minutes and accuracy on the attention-switching task at 90 minutes. The combination improved both the speed and the accuracy of attention switching at 60 minutes and reduced susceptibility to distracting information in the memory task at both 60 and 90 minutes Owen 2008. The abstract reports no alertness result for the combination arm at all — that is silence, not a reported null, and it should not be read as a failure. The authors' reading is positive: they present their results as replicating previous evidence that the combination benefits performance on cognitively demanding tasks. Worth knowing when weighing it: the work was carried out at Unilever Research and Development, a tea-industry laboratory.

The stress side comes from smaller physiological work. In Kimura 2007, twelve participants completed four counterbalanced double-blind trials around a mental arithmetic stressor, and L-theanine intake reduced heart rate and salivary immunoglobulin A responses relative to placebo, with heart-rate-variability analyses indicating those reductions were likely attributable to attenuated sympathetic activation Kimura 2007. The authors phrase the underlying mechanism as a suggestion rather than a demonstration. Twelve people is a very small sample, the outcomes are physiological responses to a laboratory task rather than a day's mood, the abstract does not state the theanine dose, and one of the co-authors is affiliated with a commercial theanine producer.

The Dose Gap

Here is where the popular version parts company with the papers. Neither 2008 trial was a study of matcha. They dosed two molecules that tea happens to contain, in a single session, in healthy volunteers — and the theanine doses they used are the crux.

The figure that circulates online comes from the composition review, and it is routinely misread. Kochman 2020 reports that Unno et al. found as much as 44.65 mg of theanine per gram in matcha tea samples — an upper observation from one sample set, not a typical value — and in the same passage reports a far lower figure from a different source, 6.1 mg per litre in matcha tea infusions. Unlike the caffeine figures, no range is offered for theanine in the powder, so there is no central number in that review to build an estimate on. Anyone converting 44.65 mg/g into a per-cup dose is using the highest value in the literature the review surveyed.

The measured trial products settle it better. The 2070 mg daily matcha dose in Baba 2021 supplied 48.0 mg of theanine, the 2.7 g daily dose in the 2024 sleep trial contained 50.3 mg Baba 2024, and the 2 g dose given daily for a year contained 48.1 mg Uchida 2024 — three separately analysed products landing in the same narrow band. So a real serving in the 2 to 2.7 g range delivered roughly half the 100 mg Owen 2008 used and about a fifth of the 250 mg in Haskell 2008. On those measured concentrations, reaching Haskell's theanine dose from powder alone would take something in the region of 10 to 13 grams in one cup, which at the caffeine concentrations measured in those same products would carry far more caffeine than either trial used. If the true theanine concentration in the tin in your cupboard is lower than the trial products', the gap widens rather than narrows.

That is the pinch. Serve matcha at a normal strength and you fall well short of the theanine doses that produced the attention effects. Serve it strongly enough to reach them and you are no longer drinking the modest-caffeine beverage that made the switch attractive in the first place.

The Trials That Used Real Matcha

The obvious objection to all of the above is that someone should test matcha itself, for longer than one sitting. Three of the trials in this reference list did — a sample of that literature rather than all of it.

Baba 2021 was a 12-week double-blind, randomised, placebo-controlled, parallel-group study in 51 healthy Japanese men and women aged 50 to 69, with three arms — placebo, isolated caffeine at 66.6 mg, and matcha supplying 66.2 mg of caffeine plus 48.0 mg of theanine and 170.0 mg of catechins — examining cognitive function related to mild acute psychological stress. The authors conclude that ingesting matcha with caffeine improves both attention and work performance when suffering from psychological stress compared with caffeine alone Baba 2021. They also note that a reduced reaction time observed after a single dose of matcha was likely due to caffeine, which is a useful piece of internal candour: the acute speed effect looked like the caffeine, and the matcha-specific difference showed up under stress.

The second trial is largely a null. Over 4 weeks, 46 healthy Japanese adults aged 27 to 64, 23 per group, taking 2.7 g of matcha daily showed no significant differences from control in total sleep time, sleep latency, wake after sleep onset or sleep efficiency measured by electroencephalography; the one EEG measure that did move was time spent in bed after waking, which was shorter with matcha than with placebo at p below 0.05. Changes in satisfaction with sleep time and in Beck Depression Inventory-II scores were reported only as a trend, at p below 0.1 Baba 2024. The authors titled the paper around matcha possibly enhancing mental well-being, and on the four sleep parameters the trial was built around, objective sleep architecture did not move.

The third is the longest and largest of them, and it is the one usually missing from the matcha argument. Uchida 2024 was a 12-month double-blind, randomised, placebo-controlled trial in 99 Japanese adults aged 60 to 85 with cognitive decline, 91 of whom were analysed, taking 2 g of matcha daily in capsules. Its two primary endpoints, the Japanese Montreal Cognitive Assessment and an activities-of-daily-living scale for mild cognitive impairment, did not differ from placebo. One secondary outcome did: perception of facial emotion improved with matcha, at p = 0.028. Sleep there was measured by questionnaire rather than EEG, and the difference in Pittsburgh Sleep Quality Index scores did not reach significance, at p = 0.088 Uchida 2024. A year of daily matcha in an older, cognitively declining population, in other words, moved one social-cognition measure and left the headline cognitive and sleep endpoints where they were. Its capsule dose also happens to corroborate the arithmetic above from a third laboratory: 2 g of a defined matcha supplying 48.1 mg of theanine.

All three were run in Japanese adults, each used a fixed daily dose of a defined product, and none compared matcha with coffee. They are the closest thing in this reference list to a test of the drink rather than the molecules, and what they deliver is small, conditional and largely null — not the transformation the category is sold on.

How Load-Bearing The Evidence Is

Pooled analyses give a sense of scale. A 2025 systematic review and meta-analysis of tea, L-theanine, or L-theanine plus caffeine in healthy participants, searching to August 2023, included 50 randomised controlled trials in the review, of which 15 were eligible for at least one meta-analysis Payne 2025. The pooled effects for theanine plus caffeine versus placebo were small and mixed in significance: choice reaction time gave a standardised mean difference of −0.48 with a 95% confidence interval of −1.01 to 0.05, which crosses zero, while digit vigilance task accuracy gave 0.20, 95% CI 0.02 to 0.38 Payne 2025. Statistically significant, but small.

For theanine on its own, a PROSPERO-registered 2025 meta-analysis pooled five randomised controlled trials involving 148 healthy adults, and its authors titled the paper, in as many words, promising but not completely conclusive Mátyus 2025. That is a fair description of the base the calm-focus claim stands on: real, partly replicated, modest effects on specific laboratory tasks, pooled from a small subset of the trials that have been run.

Why Two People React Differently To The Same Cup

The complaint that sends people to matcha is jitters, and the most consistent explanations for it have little to do with tea. Nehlig 2018, a pharmacology review, describes how widely caffeine handling varies between individuals: pharmacokinetics are highly variable because of polymorphism at the CYP1A2 isoform of cytochrome P450, which metabolises 95% of ingested caffeine, plus polymorphism at N-acetyltransferase 2, and there are several polymorphisms at caffeine's main brain target, the adenosine A2A receptor. Genetic studies, including genome-wide association studies, have identified loci involved in caffeine consumption and its consequences for sleep and anxiety Nehlig 2018. The same review lists age, sex and hormones, liver disease, obesity, smoking and diet as factors affecting metabolism and clearance, and notes that caffeine interacts with many medications.

Nehlig also observes that most individuals adjust their intake according to the effects they experience, and that the quantity needed to reach a desired effect varies widely between people Nehlig 2018. Being a review, it cannot tell you that your own jitters are genetic, and it makes no claim about matcha or about switching beverages. What it does establish is that two people drinking the same espresso are not necessarily having the same pharmacological experience. Habituation is a separate variable that changes what any given dose feels like, and we cover it in our piece on how caffeine tolerance builds.

There is a cheap experiment here that the trials cannot run for you. Weigh the powder, count the cups, and total the milligrams for a week to see whether your intake actually fell when you switched. A drop in total dose and a change in beverage chemistry are two different explanations for feeling better, and only one of them is measurable on a kitchen scale.

The Long-Run Health Comparison

If the question is instead which drink is better over decades, the evidence is observational on both sides and does not resolve into a ranking. Poole 2017 is an umbrella review of other people's meta-analyses — 201 meta-analyses of observational research covering 67 unique health outcomes, plus 17 meta-analyses of interventional research covering nine outcomes. It found coffee consumption more often associated with benefit than harm, and its summary estimates showed the greatest risk reduction at intakes of three to four cups a day versus none: all-cause mortality relative risk 0.83 (95% CI 0.79 to 0.88), cardiovascular mortality 0.81 (0.72 to 0.90) and cardiovascular disease 0.85 (0.80 to 0.90) Poole 2017. Harmful associations were largely nullified by adequate adjustment for smoking, except in pregnancy, where high versus low or no consumption was associated with low birth weight (odds ratio 1.31, 1.03 to 1.67), pregnancy loss (1.46, 1.06 to 1.99) and preterm birth in the second trimester (1.12, 1.02 to 1.22); the first-trimester preterm interval, 1.22 (1.00 to 1.49), touches 1.00. Fracture risk was associated with coffee drinking in women but not in men. The authors' own position is notably unalarmed: they conclude coffee seems generally safe within usual levels of intake, call for randomised trials to establish causality, and suggest that outside pregnancy coffee could be tested as an intervention without significant risk of harm, with women at increased fracture risk possibly excluded Poole 2017. An erratum to that paper was published in 2018.

For green tea, the Ohsaki cohort followed 40,530 Japanese adults aged 40 to 79 with no history of stroke, coronary heart disease or cancer at baseline, for up to 11 years for all-cause mortality and up to 7 years for cause-specific mortality Kuriyama 2006. Green tea consumption was inversely associated with all-cause and cardiovascular mortality, and the association with all-cause mortality was stronger in women (P = .03 for the interaction with sex). In women, all-cause hazard ratios across rising consumption were 1.00, 0.98 (0.84 to 1.15), 0.82 (0.70 to 0.95) and 0.77 (0.67 to 0.89), P below .001 for trend; in men, only the five-or-more-cups-a-day category reached significance at 0.88 (0.79 to 0.98), with the 1–2 and 3–4 cup intervals both crossing 1.00 Kuriyama 2006. Cancer mortality hazard ratios were not significantly different from 1.00 in any consumption category. The exposure was self-reported cups of green tea per day in Japan — not matcha, and with no measurement of caffeine or theanine intake.

These are separate literatures, in different populations, with different confounders and different exposures. They do not stack into a head-to-head verdict, and choosing between the two drinks on longevity grounds means choosing on evidence that was never designed to answer that question.

How To Decide For Yourself

Stop treating this as a contest between two drinks and treat caffeine as a dose. If you get anxious, wired or wakeful at night, the number to look at is your total daily milligrams against the intakes EFSA 2015 describes as not raising safety concerns, on the per-kilogram basis EFSA uses, rather than which plant supplied them.

If you have already cut the dose and still notice a difference, the combination trials give you something real but narrow to hang it on: changes in specific reaction-time, working-memory and attention-switching measures in single sessions Haskell 2008 Owen 2008, at theanine doses roughly two to five times what a measured 2 to 2.7 g matcha serving supplied in the trials that analysed their product Baba 2021. Where matcha itself has been fed for weeks or months in the trials cited here, the reported advantage over caffeine alone was specific to performance under mild psychological stress Baba 2021, and the sleep evidence is thin rather than absent: four EEG measures unchanged over 4 weeks, with time in bed after waking shorter on matcha Baba 2024, and a Pittsburgh Sleep Quality Index difference over 12 months that did not reach significance Uchida 2024.

And if you simply prefer the taste, that is a complete reason. It requires no evidence base at all.

Frequently asked questions

Does matcha have more or less caffeine than coffee?

It depends on the serving. Your dose is powder mass multiplied by concentration, and Kochman 2020 quotes 18.9 to 44.4 mg of caffeine per gram for matcha against 11.3 to 24.67 mg/g for green teas generally, ranges that overlap. Measured trial products are firmer: 2070 mg of matcha daily supplied 66.2 mg of caffeine in Baba 2021, and 2.7 g contained 71.5 mg in the 2024 sleep trial. EFSA's caffeine opinion assumed 90 mg per 150 mL cup only to back-calculate doses in studies that did not report cup size, and it notes coffee concentrations vary with beans, process and preparation, so that is a scale marker rather than a fact about your cup.

Is the calm alertness from matcha real?

There is genuine trial support for the ingredient combination, though not for matcha as the vehicle. In Haskell 2008, caffeine alone gave faster digit vigilance reaction time, better Rapid Visual Information Processing accuracy and less self-reported mental fatigue, while the combination with 250 mg of L-theanine reproduced those and added faster simple reaction time, faster numeric working memory reaction time and better sentence verification accuracy, with lower headache and tired ratings and higher alert ratings; L-theanine alone increased headache ratings and reduced correct serial sevens. Only delayed word recognition reaction time is reported there as a significant interaction between the two compounds. In Owen 2008, caffeine improved alertness at 60 minutes and attention-switching accuracy at 90 minutes, and adding 100 mg of L-theanine improved attention-switching speed and accuracy and reduced distractibility; that abstract reports no alertness result for the combination arm either way. Both were acute, single-session studies of isolated compounds.

How much L-theanine is in a serving of matcha?

Less than those trials used. The 2070 mg daily matcha dose in Baba 2021 supplied 48.0 mg of theanine, and 2.7 g in the 2024 sleep trial contained 50.3 mg — roughly half the 100 mg in Owen 2008 and about a fifth of the 250 mg in Haskell 2008. The 44.65 mg per gram figure that circulates online is reported in Kochman 2020 as an upper observation from one sample set, quoted alongside a much lower 6.1 mg per litre for infusions, with no typical value given.

Will switching from coffee to matcha stop my jitters?

The evidence cited in this article does not include any trial of switching beverages, so it cannot answer that. What it can say is that Nehlig 2018 describes large inherited and acquired differences in caffeine metabolism and response, including polymorphisms at CYP1A2 and at the adenosine A2A receptor and loci linked to caffeine's effects on sleep and anxiety. If the switch also lowers your total daily caffeine, a change in dose and a change in beverage chemistry become two explanations this evidence cannot separate.

Is matcha healthier than coffee for long-term health?

The evidence does not settle it. Poole 2017, an umbrella review of meta-analyses, found coffee more often associated with benefit than harm, with all-cause mortality relative risk 0.83 (0.79 to 0.88) at three to four cups a day versus none, and exceptions in pregnancy and for fracture risk in women; its authors judged coffee generally safe at usual intakes and called for randomised trials. Kuriyama 2006 found green tea drinking associated with lower all-cause and cardiovascular mortality in 40,530 Japanese adults, with no significant association for cancer mortality. These are separate observational literatures in different populations, and the green tea exposure was cups of tea, not matcha.

Should I drink matcha before training instead of coffee?

For caffeine's purposes, the dose and its timing matter more than the plant, which is the subject of our separate piece on caffeine timing. Matcha can supply caffeine perfectly well, and the trials that analysed their product put a 2 to 2.7 g serving in the high sixties to low seventies of milligrams Baba 2021. What no source cited here supports is a claim about which drink gives more predictable dosing — EFSA 2015 notes coffee's own caffeine content varies with beans, process and preparation. If you need a known dose, weigh or measure whatever you use.

References