In plain words

In most trials, people who took L-theanine alone scored no better on thinking tests than people given a placebo. One trial found faster reactions, and another found a possible benefit from a product called LGNC-07. One found fewer correct answers on a subtraction test.

The short answer

Evidence leans to no effect of L-theanine alone. Four of the seven trials found no difference from placebo: 200 mg daily for 4 weeks in 30 adults; 50 mg in 24 adults (no duration in the ledger); one 200 mg drink (no n in the ledger); 12 weeks in 69 people (no dose in the ledger). One 250 mg dose in 24 adults lowered correct subtractions. Two found gains: faster reactions after one 100.6 mg dose in 50 people, and a 16-week trial of 1,680 mg LGNC-07 in 91 people that suggests potential. Certainty is low, because the trials are small, mostly single-dose and use different tests.

Where the evidence sits

Better than placebo

■ Randomized trial, Mátyus et al., 2025 · 50 people · single dose
■ Randomized trial, Park et al., 2011 · 91 people · 16 weeks

No difference from placebo

■ Randomized trial, Hidese et al., 2019 · 30 people · four weeks · also reported by Alzheimer's Drug Discovery Foundation, 2025 · scores rose from baseline
■ Randomized trial, Dodd et al., 2015 · 24 people
■ Randomized trial, White et al., 2016 · ledger has no n · single dose
◪ Pooled analysis, Mátyus et al., 2025 · may overlap studies counted here · faster at higher doses
□ Randomized trial reported by Alzheimer's Drug Discovery Foundation, 2025 · 12-week trial ages 50 to 69 · no new data

Worse than placebo

■ Randomized trial, Haskell et al., 2007 · ledger has no n · acute (single dose)

■ measured by the source's own study · ◪ pooled analysis, may overlap studies counted here · □ reported, reviewed or argued, no new data

Evidence weight7070 / 100?

Who says what

SourceFindingDesignPeopleDose and lengthFunding and independence
Hidese, 2019No differenceRandomized30200 mg, 4 weeksNot in ledger
Dodd, 2015No effectRandomized2450 mgNot in ledger
Haskell, 2007Fewer correct subtractionsCrossover24250 mg onceNot in ledger
White, 2016No changeRandomizedNot in ledger200 mg onceNot in ledger
Park, 2011Suggests potentialRandomized, LGNC-07911,680 mg, 16 weeksNot in ledger
Mátyus, 2025Mixed reaction timesPooled; may include above148Single dosesNot in ledger
Alzheimer's Drug Discovery Foundation, 2025No effectFoundation describing trials30; 694; 12 weeksNot in ledger
Hidese, 2019

No difference

Design
Randomized
People
30
Dose and length
200 mg, 4 weeks
Funding
Not in ledger
Dodd, 2015

No effect

Design
Randomized
People
24
Dose and length
50 mg
Funding
Not in ledger
Haskell, 2007

Fewer correct subtractions

Design
Crossover
People
24
Dose and length
250 mg once
Funding
Not in ledger
White, 2016

No change

Design
Randomized
People
Not in ledger
Dose and length
200 mg once
Funding
Not in ledger
Park, 2011

Suggests potential

Design
Randomized, LGNC-07
People
91
Dose and length
1,680 mg, 16 weeks
Funding
Not in ledger
Mátyus, 2025

Mixed reaction times

Design
Pooled; may include above
People
148
Dose and length
Single doses
Funding
Not in ledger

Did one dose change test scores?

White and colleagues (2016) gave a drink with 200 mg of L-theanine or a placebo once. The ledger has no n for this trial:

No treatment-related cognitive performance changes were observed.

Dodd and colleagues (2015) found no effect of 50 mg alone in 24 adults, and the ledger has no duration for it. After one 250 mg dose in 24 adults, Haskell and colleagues (2007) recorded fewer correct serial seven subtractions. Mátyus and colleagues (2025) describe a trial by Baba and colleagues in which 50 people given one 100.6 mg dose had faster simple reaction times than the placebo group. Their pooled estimate may include the trials above. It found that single doses of 50 to 400 mg did not change simple reaction time in 126 adults, while higher single doses up to 400 mg sped two other reaction tests in 148 adults.

Did weeks of daily use do better?

Hidese and colleagues (2019) gave 30 healthy adults 200 mg a day or a placebo for four weeks. Verbal fluency and executive function scores rose from baseline, but not against placebo:

When score changes were compared between the L-theanine and placebo administrations, no significant difference was found for any cognitive function.

The Alzheimer's Drug Discovery Foundation (2025) describes a four-week trial in 30 middle-aged people with no difference on 12 cognitive outcomes, which appears to describe the same trial. It also describes a 12-week trial in 69 people aged 50 to 69 with no significant effect. The ledger has no dose for either. Park and colleagues (2011) gave 91 people 1,680 mg of a product named LGNC-07 for 16 weeks and wrote that their study suggests it has potential for cognitive improvement.

Why the sources disagree

Dose, length (one dose or several weeks) and test (reaction time to verbal fluency) differ across studies. Hidese's scores rose from baseline but not against placebo. Park tested LGNC-07. Baba's trial and the 12-week trial reach the ledger only through other sources. Mátyus's pooled estimate may include the trials above.

What would settle it

An independent placebo-controlled trial of L-theanine alone at 200 mg a day for 12 weeks in at least 200 healthy adults would settle it. The trial would register its cognitive tests and main outcome in advance and report results against placebo rather than baseline. A matching result from a second research group would raise certainty from low to moderate.

How this page was made

The 7 sources are five studies of L-theanine and cognition, a 2025 meta-analysis that may include them, and the Alzheimer's Drug Discovery Foundation's 2025 evidence rating, found through a medical literature search. Studies in people with schizophrenia, depression, anxiety disorder or ADHD were left out. Every claim was copied word for word into the Longevity Circle ledger, checked against the original text and approved by a person. Certainty follows GRADE levels. Drafting was assisted by AI.

Changelog: 30 Sep 2026, first version, 7 sources.