The Cortisol Story Is More Complicated Than the Headlines

September 11, 2026

"Lowers cortisol" is one of the most repeated claims in stress-supplement marketing — and one of the least often checked against an actual trial. We went and checked. Then we checked our own article, and found we'd made one of the same mistakes we were criticising.

Correction notice (September 2026): Five corrections, and one of them is exactly the error this article was written to call out. (1) We reported that ashwagandha "serum cortisol dropped by 27.9% versus placebo." It did not. 27.9% is the fall from that arm's own baseline; the placebo arm fell 7.9% from its own baseline. The between-group comparison was significant, but the placebo-adjusted gap is roughly 20 percentage points, not 27.9% — presenting a within-group change as a placebo-controlled difference is the single most common way cortisol claims get inflated, and we did it. (2) We said the four-week L-theanine trial showed improvements in "self-rated stress and anxiety"; the significant findings were depression and trait anxiety scores, and we omitted that two cognitive tests improved only on placebo and a third showed nothing. (3) We described the L-theanine-plus-caffeine trial as showing the combination beat "either ingredient taken alone." That trial compared the combination with placebo only, and three of its cognitive outcomes were null. (4) Sample size and product details for the acute-stress trial were wrong (34 completers, and an L-theanine-based nutrient drink, not isolated L-theanine). (5) We wrote that 200mg/day of L-theanine is "the range we dose to." It isn't, and we cannot show that it is — see the section at the end, which replaces that claim.

The trial

A randomized, double-blind, placebo-controlled crossover trial gave 30 healthy adults (9 men, 21 women; mean age 48.3) 200mg/day of L-theanine for four weeks, then measured stress-related symptoms, sleep quality, cognitive function — and cortisol, in both saliva and serum.¹

What actually improved — and what didn't

Several outcomes moved, and it's worth naming which ones rather than rounding them into "stress." Significant versus placebo: depression scores on the Self-rating Depression Scale (p=0.019), trait anxiety on the State-Trait Anxiety Inventory (p=0.006), sleep quality on the Pittsburgh Sleep Quality Index (p=0.013), and executive function (p=0.031). Verbal and letter fluency also improved (p=0.001–0.002), though notably most in a lower-performing subgroup — a subgroup finding, not a whole-sample one.¹

And the nulls, which an earlier version of this article skipped: Trail Making Test A and B improved only in the placebo condition, and the Stroop test showed no significant benefit. A trial where some cognitive tests go the other way is not a clean win, and saying so is the whole point of this article.

What didn't move at all

Cortisol. In the authors' words, "four weeks L-theanine administration had no significant effect on cortisol or immunoglobulin A levels in the saliva or serum."¹ Despite being the most common claim attached to stress supplements generally, this trial found nothing on it — not a small effect, no significant effect at all, in either measurement medium.

How L-theanine is thought to work

L-theanine is an amino acid found almost exclusively in tea leaves, and most of its plausible mechanism traces back to a simple structural fact: it closely resembles glutamate, the brain's main excitatory neurotransmitter. That resemblance lets it cross the blood-brain barrier and interact with glutamate receptors, generally in a direction that softens excitatory signaling rather than amplifying it, alongside smaller effects on GABA and dopamine activity.² The clearest human evidence for this comes from EEG research: a placebo-controlled study found L-theanine measurably increased alpha-band brain wave activity during a sustained visuo-spatial attention task — a pattern associated with what researchers describe as "relaxed alertness," distinct from the drowsiness a sedative produces.² Separately, a small crossover trial found 200mg of L-theanine blunted heart rate and salivary immune-marker (IgA) responses to an acute mental-arithmetic stressor, with the authors attributing the effect to "inhibition of cortical neuron excitation" — a mechanistic story consistent with the glutamate-receptor explanation, even though that particular study never measured cortisol at all.³ Our longer write-up of the L-theanine evidence goes through the mechanism in more detail.

Stress isn't one biomarker — it's a set of loosely connected systems. A supplement can shift how you feel and think without laying a hand on any one of them.

Why this distinction matters

Cortisol is a hormone, not a feeling — it's tempting to use it as a proxy for "biological proof" that a stress intervention worked. This trial is a useful reminder that a real, statistically significant improvement in how someone feels and functions doesn't require a matching change in a specific biomarker. The subjective benefit here was real. The cortisol mechanism, at least in this trial, wasn't.

Is this a one-off, or a pattern?

Thirty people is not nothing — a crossover design is efficient precisely because each participant serves as their own control — but it's also not the whole literature. The honest next question is whether other L-theanine trials that measured cortisol found the same null result, or something different.

The answer is mixed, which is itself informative. A separate randomized, double-blind, placebo-controlled crossover trial gave 34 healthy adults aged 18–40 an L-theanine-based nutrient drink containing 200mg of L-theanine, then exposed them to a standardized mental-arithmetic stressor while tracking salivary cortisol. That trial did find a significant effect — but a narrower one than "lowers cortisol" implies: the cortisol response to the stressor was lower on active treatment three hours post-dose (z=−1.98, p=0.047), with no significant difference at one hour (z=1.28, p>0.05).⁴ Two caveats: p=0.047 is a result clearing the threshold by a hair, and the intervention was a multi-ingredient drink rather than isolated L-theanine, so the effect cannot be cleanly assigned. It's also a claim about how sharply cortisol spikes in response to an acute stressor in one time window — not whether resting cortisol trends downward over weeks of daily use, which is what the four-week trial above measured and didn't find.

Zooming out, a 2026 systematic review and meta-analysis in Molecular Psychiatry pooled 31 randomized controlled trials of L-theanine covering 1,168 participants.⁵ Its findings are worth reading precisely:

Attention: a single 200mg dose taken 30–60 minutes before testing significantly improved choice reaction time (SMD 0.51; 95% CI 0.25–0.77) — the most robust finding in the review.
Acute stress: a modest effect (SMD 0.31), which the authors say was "largely influenced by studies with a high risk of bias."
Anxiety: effects were inconsistent and statistically non-significant, apart from a single trial in psychotic anxiety. This is worth sitting with, because the four-week trial above found a trait-anxiety benefit — and the pooled literature does not support generalising it.
Depressive symptoms: a significant reduction (SMD 0.69; 95% CI 0.13–1.25) after excluding one outlier, in healthy individuals on single doses.

The review's pooled outcomes were cognitive and affective; it did not produce a pooled cortisol estimate. Cortisol simply hasn't been a consistently measured outcome in this literature, and where it has been measured, the result depends heavily on exactly what's being tracked — chronic resting levels versus acute stress reactivity — and when. That's a reason for humility, not for picking the friendliest number.

How this compares to other stress-marketed ingredients

This is where L-theanine's profile becomes more interesting rather than less credible. Not every ingredient sold under the "stress support" banner works the same way, and lumping them together under one blanket claim erases real differences the trial data actually show.

Ashwagandha is the clearest contrast, and also the place where we got it wrong. A randomized, double-blind, placebo-controlled trial gave chronically stressed adults a high-concentration, full-spectrum ashwagandha root extract for 60 days.⁶ Serum cortisol in the ashwagandha arm fell 27.9% from its own baseline. In the placebo arm, cortisol also fell — by 7.9% from its own baseline. The between-group comparison was statistically significant (p=0.0006), so this is a real, trial-backed biomarker effect. But the placebo-adjusted difference is roughly 20 percentage points, not 27.9%, and "27.9% versus placebo" — the phrasing this article previously used, and the phrasing that appears across most of the internet — silently converts a within-arm change into a controlled effect. If you take one thing from this article, take that distinction, because it is the single most common way a supplement biomarker claim gets inflated without anyone technically lying.

Ashwagandha is thought to act partly by modulating the hypothalamic-pituitary-adrenal (HPA) axis — the loop governing how much cortisol the adrenal glands release — which is mechanistically different from L-theanine's glutamate-receptor and alpha-wave activity. Our piece on why the extract matters more than the herb covers why results don't transfer between ashwagandha products.

27.9% vs 7.9%

Serum cortisol fell 27.9% from baseline in the ashwagandha arm of a 60-day randomized trial — and 7.9% from baseline in the placebo arm. The between-group difference was significant (p=0.0006). The headline "27.9% cortisol reduction" quietly drops the second number.⁶

Magnesium sits in a different place again. The RCT evidence ties supplementation to small measurable improvements in insomnia severity, largely through mechanisms tied to GABA-A receptor activity and nervous-system regulation, rather than a direct, well-established cortisol effect of its own — as our pieces on what magnesium can and can't do for cortisol and what the newest magnesium sleep trial found both set out. None of this makes L-theanine a weaker ingredient because it doesn't move cortisol — it means "lowers cortisol" is a specific, testable, ingredient-by-ingredient claim. Ingredients earn their claims one trial at a time, not by category association.

Timing, pairing, and dose in practice

The dose range with the clearest RCT support for L-theanine's subjective and cognitive effects is modest: single doses around 200mg for acute attention and calm-alertness effects, and 200mg/day sustained for the four-week outcomes above.¹⁵

In practice, people use L-theanine two ways: alone, often later in the day or during a stressful stretch; or paired with caffeine, for what's often called calm focus. A randomized, double-blind, crossover trial in 44 young adults tested 97mg of L-theanine with 40mg of caffeine — roughly half a cup of coffee's worth — against placebo.⁷ It significantly improved accuracy during task switching and self-reported alertness (both p<0.01) and reduced self-reported tiredness (p<0.05). It found no significant effects on visual search, choice reaction times, or mental rotation. And it did not compare the combination against either ingredient alone — an earlier version of this article said it did, which was wrong, and it means the trial cannot tell you the pairing beats caffeine on its own. Nor did it test whether L-theanine takes the jittery edge off caffeine; that's a popular claim this study doesn't speak to. What it supports is narrower: that combination, at those amounts, helped focus attention during a demanding task compared with nothing. The full calm-focus picture is here.

Key takeaway: At 200mg/day, L-theanine's own RCT improved depression scores, trait anxiety, sleep quality and executive function — but not measured cortisol, and two cognitive tests improved only on placebo. Acute cortisol reactivity shifted in one multi-ingredient drink trial at three hours (p=0.047) but not at one hour. Ashwagandha does have trial-backed cortisol data — a 27.9% within-arm fall against 7.9% on placebo, which is a real effect and a smaller one than the headline implies. Each ingredient earns its claim from its own evidence.

What we take from this — including about our own label

We don't market L-theanine as a "cortisol lowering" ingredient, because the best available RCT data on that specific claim says it isn't one. Where cortisol is a supported claim — as it is for ashwagandha — it belongs to that ingredient specifically, at the extract and duration tested, and nowhere else.

And a correction about us, not the research. An earlier version of this article said that the 200mg/day L-theanine range "is the range we dose to." That was not a claim we can stand behind. L-theanine is one of eight ingredients inside Cognitive Support's 255mg proprietary blend, and the per-ingredient amounts are not disclosed on the label. You can do the arithmetic we should have done before writing that sentence: 255mg shared among eight ingredients cannot contain 200mg of any one of them and leave room for the other seven. We don't know the exact figure, we can't publish it, and we should not have implied our formula matches a trial dose. If 200mg of L-theanine is what you want, a single-ingredient product with the milligrams printed on the panel is the honest way to get it — and our guide to proprietary blends and underdosing explains how to run that arithmetic on anyone's label, including ours.

This article is for informational purposes only and has not been evaluated by the FDA. It is not intended to diagnose, treat, cure, or prevent any disease. Speak with a healthcare provider before starting any new supplement, especially if you take medication or have an existing health condition.

References

Hidese S, Ogawa S, Ota M, et al. Effects of L-Theanine Administration on Stress-Related Symptoms and Cognitive Functions in Healthy Adults: A Randomized Controlled Trial. Nutrients, 2019;11(10):2362. mdpi.com/2072-6643/11/10/2362
Gómez-Ramírez M, et al. (2009). The effects of L-theanine on alpha-band oscillatory brain activity during a visuo-spatial attention task. Brain Topography. pubmed.ncbi.nlm.nih.gov/18841456
Kimura K, et al. (2007). L-theanine reduces psychological and physiological stress responses. Biological Psychology. pubmed.ncbi.nlm.nih.gov/16930802
White DJ, de Klerk S, Woods W, et al. (2016). Anti-Stress, Behavioural and Magnetoencephalography Effects of an L-Theanine-Based Nutrient Drink: A Randomised, Double-Blind, Placebo-Controlled, Crossover Trial. Nutrients, 8(1):53. mdpi.com/2072-6643/8/1/53
Gerolymos C, Saddier E, Boyer L, Fond G. (2026). Cognitive and affective effects of L-Theanine: a systematic review and meta-analysis of 31 randomized trials. Molecular Psychiatry. nature.com/articles/s41380-026-03727-9
Chandrasekhar K, Kapoor J, Anishetty S. (2012). A prospective, randomized double-blind, placebo-controlled study of safety and efficacy of a high-concentration full-spectrum extract of ashwagandha root in reducing stress and anxiety in adults. Indian Journal of Psychological Medicine. pubmed.ncbi.nlm.nih.gov/23439798
Giesbrecht T, Rycroft JA, Rowson MJ, De Bruin EA. (2010). The combination of L-theanine and caffeine improves cognitive performance and increases subjective alertness. Nutritional Neuroscience, 13(6):283-290. pubmed.ncbi.nlm.nih.gov/21040626