Reading the L-Theanine Evidence: Part One: What the Cognition Trials Measured

Abstract

L-theanine is an amino acid that makes up nearly half the free amino acids in green tea. In the last five years it has accumulated systematic reviews and meta-analyses faster than almost any other compound in its category, and those reviews do not agree with each other.

This is the first of five parts working through that literature. It covers cognition: what has been measured, in whom, at what dose, and what survives when the results are pooled rather than quoted individually.

We should state our position before the evidence. L-theanine is in two MindHeaven® formulations. That is a reason to read this literature more carefully than a disinterested party would, not less, and everything below is written on the assumption that a reader will check it.

1.Why a Series

Three groups published overlapping reviews of L-theanine and cognition between 2021 and 2025. A reader looking for a verdict can find one that says the compound is a safe and effective cognitive enhancer, one that says its superiority over placebo is not proven, and one that says the effects are real but the confidence intervals frequently include nothing at all.

All three are honest pieces of work. They disagree because they asked slightly different questions of slightly different study sets, which is the ordinary way a small literature behaves and the reason single quotations from it are worth so little.

The remaining parts take sleep, stress, mechanism and the question of whether the science supports what the market says about it. This part establishes the ground: what the cognition trials actually did.

2.The Systematic Review That Reached a Verdict

Anas Sohail, Fernando Ortiz and colleagues published a systematic review in Cureus at the end of 2021, searching PubMed for trials of L-theanine with caffeine. Their conclusion: the combination is likely a safe and effective cognitive enhancer.

The review found five studies. Three of them came from the same research group — Kahathuduwa and colleagues, publishing in 2017, 2018 and 2020 — and the participant counts were 20, 9 and 5 respectively. The 2020 study, the only one in a clinical population, enrolled five boys with attention-deficit hyperactivity disorder.

The other two studies were larger in the relevant sense: Haskell and colleagues tested 250 mg of theanine with 150 mg of caffeine in 24 adults, and Baba and colleagues gave matcha capsules to 51 Japanese adults aged 50 to 69 for 12 weeks.

One result inside that review is more interesting than its conclusion. In the ADHD study, caffeine alone and L-theanine alone each made inhibitory control worse — stop-signal reaction times lengthened, with p-values of 0.031 and 0.053. Only the combination improved it. Whatever else the pairing does, the individual compounds did not simply add up.

A second, from Haskell's trial: participants given L-theanine alone reported more headache and completed fewer correct serial subtractions than on placebo. That is a negative result for the isolated compound, sitting inside a review whose conclusion is positive about the pair.

3.The Meta-Analysis of L-Theanine Alone

In October 2025 Rebeka Mátyus, Barbara Tóth and Dezső Csupor at the University of Szeged published what they describe as the first meta-analysis of L-theanine as a monotherapy for cognition. The title is unusually candid: “Promising, but Not Completely Conclusive”.

Their search returned 2,301 records and ended with five randomised placebo-controlled trials in 148 healthy adults. Doses ranged from 50 mg to roughly 500 mg. The protocol was registered in PROSPERO before the work began, and the authors note that no amendments were made to it.

The results split by task. On simple reaction time, nothing: a mean difference of −0.46 milliseconds, with a confidence interval running from −15.65 to 14.73. On the Stroop test, nothing for congruent stimuli (−37.38 ms, CI −86.39 to 11.62), and for incongruent stimuli a non-significant change in the wrong direction (109.28 ms, CI −8.72 to 227.27).

On the third measure — rapid visual information processing and recognition visual reaction time — the pooled result was significant, and only at the higher doses: −15.20 milliseconds, confidence interval −28.99 to −1.41.

Fifteen milliseconds is the size of the effect. Whether that matters outside a laboratory is a separate question the authors do not claim to answer.

There is also a result pointing the other way that deserves to be said out loud. At the highest dose tested, 400 mg, simple reaction time was non-significantly worse than placebo — a mean difference of 5.91 milliseconds. The authors write plainly that there is currently insufficient literature to explain the deterioration.

Their GRADE assessment rated the certainty of evidence across all four outcomes as low or very low, citing small samples, inconsistency and imprecision. Their own summary is that the superiority of L-theanine over a placebo in enhancing cognition is not proven.

4.The Largest Analysis, and Who Paid For It

Edward Payne, Baukje de Roos and colleagues at the Rowett Institute in Aberdeen published the broadest synthesis in Nutrition Reviews in 2025: 50 randomised trials reviewed, 15 pooled, covering tea, theanine alone, and theanine with caffeine, against cognition, mood and sleep.

The funding is declared at the foot of the paper and should be read alongside the findings. Payne's doctoral studentship is funded by Lipton Teas and Infusions. One co-author is an employee of Lipton Teas and Infusions; another is an employee of Unilever, a company that produces and markets tea.

We say this with the same qualification we applied to the industry-funded trials covered elsewhere in this section: disclosed funding is not misconduct, most nutrition research is industry-funded because nobody else pays for it, and the analysis itself is conspicuously restrained. It reports several results that a sponsor would not have chosen.

The first is the risk-of-bias assessment. Of the studies evaluated, zero were judged at low risk of bias, 12 raised some concerns, and 25 were judged at high risk. That is the state of the underlying literature, stated by the people best placed to know it.

The second is what happened to alertness. An earlier meta-analysis had found that theanine plus caffeine significantly improved alertness in both the first and second hours — but only after excluding one study as a statistical outlier. Payne's team found no clinical or methodological grounds to justify that exclusion, kept the study in, and the effect stopped being conclusive.

What did survive: attention switching accuracy in the first hour after theanine plus caffeine (standardised mean difference 0.4, CI 0.24 to 0.57), the same task in the second hour (0.33, CI 0.13 to 0.54), digit vigilance accuracy in the second hour (0.20, CI 0.02 to 0.38), and simple reaction time in the second hour (−0.71, CI −0.92 to −0.50, about 27 milliseconds). For theanine alone, choice reaction time in the first hour (−0.35, CI −0.61 to −0.10, about 18 milliseconds).

The authors' own framing of the limits is worth quoting: the confidence intervals frequently highlighted the uncertainty surrounding the direction and magnitude of these differences.

5.Where the Three Agree

Read together rather than separately, the three reviews converge on more than their conclusions suggest.

The effects, where they appear, are on attention rather than on memory or reasoning. Attention switching, digit vigilance, visual information processing, choice reaction time. Nothing here concerns learning, recall or intelligence, and the compound is not tested as though it did.

They are acute. Almost every measurement was taken in the first or second hour after a single dose, timed to match the pharmacokinetics — L-theanine crosses the blood–brain barrier within about 30 minutes and peaks in plasma between 30 minutes and two hours. The literature on what daily intake does over weeks barely exists.

They are small. Fifteen to twenty-seven milliseconds on reaction times, one to two percentage points on accuracy. Real, repeatedly measured, and not the kind of change anyone notices from the inside.

And they are stronger for the pairing than for the amino acid alone. Every analysis that separated theanine from caffeine found caffeine doing most of the visible work, which is also the position the European Food Safety Authority took when it evaluated a claim for black tea and attention: it based the assessment on caffeine content.

6.The Dose Question

One pattern runs through all three reviews and is easy to miss because each states it in passing.

In Mátyus's analysis the significant result appeared only at the higher doses; the pooled effect at 100 mg was smaller and its interval crossed zero. In Payne's, the theanine-alone analyses that reached significance were those including the 400 mg arm, and the median dose in theanine-only interventions was 200 mg. In the trials Sohail reviewed, the doses were 200 mg and 250 mg.

The doses at which anything has been measured are, in other words, well above what tea delivers. Payne's team make the point directly: the lowest dose in any pooled analysis was 50 mg, equivalent to roughly two cups of black tea, and every other study used more than 100 mg. No included study delivered the intervention as an actual tea beverage.

This is the gap the authors identify as the field's most useful next step — trials using tea, or tea-equivalent doses, in free-living participants rather than in laboratories. Until those exist, the honest description of the evidence is that it concerns supplement doses of an isolated compound, tested in a controlled room, an hour after swallowing it.

Editorial Comment

L-theanine has no authorised health claim in the European Union. We make no claim that it improves attention, cognition or anything else, and this article should not be read as one.

We have a commercial interest in this literature and the reader should weigh what follows accordingly. L-theanine is in Clarity, and in Boost at 250 mg alongside 100 mg of caffeine. Those numbers were chosen from the trials described above, before this series was written, and we have set out the reasoning on the formulation pages. Reading these three reviews has not changed them and would not have justified changing them.

What it did change is how we would describe them. An effect of fifteen to twenty-seven milliseconds on a laboratory attention task, measured acutely, at low certainty of evidence, in a field where no study was judged at low risk of bias, is a reasonable basis for a design decision. It is not a basis for a promise, and anyone selling this compound as a route to focus or clarity is describing something the evidence does not contain.

Part Two turns to sleep, where two systematic reviews published within a year of each other reached different conclusions from an overlapping set of trials.