L-Tyrosine: The 10-Gram Nootropic Sold by the Half-Gram

Natural Adderall, Now With a Death Warning

L-tyrosine has had a strange year. For most of the last decade it lived a quiet life as the "dopamine precursor" in every focus stack on the internet: a 500 mg capsule, usually taken with caffeine, sold as the thing that gets you through a deadline without a prescription. Andrew Huberman has said he takes 500 mg of it with Alpha-GPC and coffee before mental work.¹ TikTok calls it "natural Adderall." Reddit's ADHD forums have argued about it for years. The ingredient market is worth somewhere in the low hundreds of millions of dollars, which is respectable for an amino acid you also get from cheese.²

Then, in February 2026, ScienceDaily ran a headline: "Popular brain supplement linked to shorter lifespan in men." In June, it ran the same headline again. On September 4, it ran it a third time. Each time the source was the same press release from the same publisher about the same paper, a UK Biobank analysis published in the journal Aging in October 2025 that concluded higher blood tyrosine might cost men "nearly one year" of life.^3,4

So the focus supplement is now, depending on which feed you read, either a cognitive enhancer or a slow poison. I read the trials behind the first claim and the paper behind the second. Neither story survives contact with its own data. What follows is why a single molecule can be simultaneously oversold and over-feared, and what the evidence actually supports, which turns out to be a much narrower and more interesting thing than either camp is telling you.

A Precursor Only Matters When the Factory Is Running Out

Tyrosine is a non-essential amino acid. Your liver makes it from phenylalanine, and a mixed diet delivers 2 to 4 grams a day on top of that, mostly from meat, fish, eggs, soy, and dairy.⁵ It has a lot of jobs: it is a building block for protein, the starting material for thyroid hormone, the substrate tyrosinase turns into melanin in your skin, and, the reason anyone sells it in capsules, the first step in the pathway that makes dopamine, norepinephrine, and epinephrine.

The pathway runs tyrosine → L-DOPA → dopamine → norepinephrine. The first step, catalyzed by tyrosine hydroxylase, is the rate-limiting one, and under normal conditions the enzyme is nearly saturated with tyrosine. That is the crucial detail. Under normal conditions, adding more tyrosine does not make more dopamine, for the same reason adding more flour to a bakery that already has flour does not make more bread. The bottleneck is the oven, not the ingredient.

The exception, and the entire scientific case for supplementation, is the depletion state. When catecholamine neurons fire hard and continuously, as they do under acute cold, hypoxia, or prolonged wakefulness, tyrosine hydroxylase is phosphorylated, its affinity for tyrosine drops, and precursor supply can become limiting.^6,7 In that specific window, more tyrosine may keep norepinephrine synthesis from falling behind demand. This is the "precursor loading" hypothesis, developed at MIT by Richard Wurtman's lab in the 1980s, and it makes a testable prediction: tyrosine should help when neurons are being pushed to exhaustion, and should do nothing when they are not.

That prediction, as we are about to see, is almost exactly what forty years of trials have found. The supplement industry sells the first half of the sentence and drops the second.

Where It Works: 7 to 10 Grams, Under Duress, in Uniform

The positive tyrosine literature was largely written by the U.S. military. The Army Research Institute of Environmental Medicine, the Naval Medical Research Institute, and the Air Force all ran small trials in the 1980s and 1990s asking whether tyrosine could protect cognition in soldiers under environmental stress. The doses were enormous by consumer standards, 100 to 150 mg per kilogram of body weight, which for a 70 kg adult is 7 to 10.5 grams, or 14 to 21 of the capsules you can buy.^6,8

The pattern repeats across the set. Neri and colleagues kept about twenty subjects awake overnight and found 150 mg/kg reduced psychomotor decline and vigilance lapses for roughly three hours before the effect faded.⁹ Magill's group put sleep-deprived young men through a head-to-head of tyrosine, 300 mg of caffeine, phentermine, and 20 mg of d-amphetamine, and tyrosine improved some measures, though the authors concluded only that it "may deserve further testing."¹⁰ Deijen gave 21 Dutch cadets 2 grams a day for five days of combat training and saw better memory and tracking performance, but the tyrosine came in a 42-gram protein drink, so the comparison is not clean.¹¹

Two systematic reviews in 2015 read this literature the same way I do. Jongkees and colleagues concluded tyrosine "seems to have a beneficial effect only in situations that stimulate neurotransmitter synthesis," and that its potential for treating clinical disorders "seems limited."¹² Hase and colleagues found it counteracts working-memory decrements under demanding conditions and does nothing for exercise performance.¹³ Neither review could pool effect sizes, because the studies are too small and too different.

So here is the honest summary of the "for" case: at 7 to 10 grams, in acute physiological emergencies, in trials of 8 to 23 people run mostly by the same few military labs, tyrosine blunts some of the cognitive damage. No trial ran longer than five days. The total number of humans across the entire positive literature is in the low hundreds. That is a real signal. It is also not the product being sold.

Where It Doesn't: Baseline, ADHD, Depression, and Anyone Over 60

The obvious next question is whether tyrosine helps when you are warm, rested, and at sea level, which describes the conditions under which essentially every consumer takes it. The military asked that too. Lieberman's group gave healthy, unstressed men 100 mg/kg and measured no behavioral effect at all.⁶ This was the prediction of the precursor hypothesis, and it held.

The non-stressed studies that did report benefits come almost entirely from one group at Leiden University. Colzato and colleagues gave 22 female undergraduates a single 2-gram dose in orange juice and found improvement on the 2-back working-memory task but not the 1-back.¹⁴ Steenbergen gave 22 people 2 grams and saw reduced task-switching costs.¹⁵ These are the studies cited by nearly every "dopamine precursor" product listing. They are one afternoon, one dose, 22 people, and nobody has replicated them at scale.

Bloemendaal's group confirmed the age problem a year later with a placebo-controlled fMRI study of adults aged 61 to 72. At 150 mg/kg there was no group-level benefit on a stop-signal task, and the older the participant, the more detrimental the effect on proactive slowing and on fronto-striatal signaling.¹⁷ The likely explanation is an inverted-U: dopamine signaling has an optimum, older brains sit at a different point on the curve, and pushing precursor into an already-adequate system moves them the wrong way.

The clinical claims fare worse. The only ADHD trial of tyrosine alone is an open-label study of 12 adults from 1987. Eight responded at two weeks; all eight had developed tolerance by six weeks, and the authors concluded it "is not useful."¹⁸ No placebo-controlled ADHD trial of tyrosine has been run since. For depression, Gelenberg's 1990 randomized trial gave 65 outpatients tyrosine at 100 mg/kg per day, imipramine, or placebo for four weeks; tyrosine raised a norepinephrine metabolite in the blood and had no antidepressant effect.¹⁹

Even the one legitimate medical indication is shakier than you would think. Patients with phenylketonuria cannot make tyrosine from phenylalanine, so supplementing it is standard practice. Yet the Cochrane review found only three trials worth including, 56 participants in total, and concluded there is "no evidence to suggest that tyrosine should be routinely added to the diet" beyond what is already in the medical formula.²⁰ When the European Food Safety Authority assessed L-tyrosine claims for "increased attention" and catecholamine synthesis in 2011, it did not substantiate them.²¹

One Cohort, Three Press Releases, and a Number From the Wrong Column

Now the paper that produced the headlines. Zhao, Sun, Zhang, and Ye, from the University of Hong Kong and the University of Georgia, took 272,475 UK Biobank participants who had a single fasting blood sample run through NMR metabolomics between 2006 and 2010, and followed them for a median of 11.1 years, during which 23,964 died.³ The work was funded by the NIH and the Hong Kong government, and the authors declare no conflicts. I want to be fair to it: this is a large, competently executed analysis, and the authors list their own limitations candidly. The problem is the distance between what they found and what was reported.

Let me unpack that, because it is where the story lives. The observational finding is a hazard ratio of 1.02. That is a 2% relative increase in mortality per standard deviation of a single blood measurement, taken once, and it is in the same range as measurement noise in a cohort this size. Phenylalanine, the other amino acid in the paper, had a stronger observational signal, 1.04 in both sexes, but nobody wrote headlines about phenylalanine because in the genetic arm it pointed the other way, toward longer life in men.³ The instruments, in other words, are not stable.

The "nearly one year" number is not a measurement of anyone's lifespan. It is a model output: Mendelian randomization takes genetic variants that raise blood tyrosine a little for life, checks whether people carrying them have parents who died earlier, converts a log hazard ratio into "years" by a scaling rule, and doubles it to correct for the fact that a child carries only half the parental genotype. Every step is defensible. Every step also adds assumptions, and the total is a number with a confidence interval that runs from 1.4 years lost to 0.2 years lost in the primary model and from 2.2 years lost to 1.7 years gained when the instruments came from a different population.

The sex story is the weakest part. In the observational data, the male-female difference was not statistically significant. In the univariable genetic analysis after pruning confounded variants, the estimates were −0.69 years for men and −0.68 years for women, essentially identical.³ In the multivariable model the women's IVW estimate was significant at p = 0.03. The "men only" framing exists because the abstract quotes the MR-Egger column, in which the women's estimate happens to cross zero. The authors' own methods say IVW should be primary when the Egger intercept shows no pleiotropy, which it did not (p = 0.61 in men, 0.15 in women). By their own rules, the answer is "modest, in both sexes, if it is real at all."

Then there is the matter of what the genetic instruments actually are. I went through the supplementary tables. The 74 tyrosine variants include HNF4A and HNF1A (monogenic diabetes genes), PNPLA3 and TM6SF2 (fatty liver), FTO (obesity), TRIB1 and MLXIPL (lipids), CDKAL1 and JAZF1 (type 2 diabetes), INSR (the insulin receptor), and a variant in the APOE region.³ One of the two "outliers" removed to make the external replication reach p = 0.03 was rs429358, the variant that defines APOE ε4, the strongest known human longevity locus. These are not variants that quietly raise tyrosine and do nothing else. They are variants that cause metabolic and liver disease, which raises tyrosine as a byproduct. Mendelian randomization cannot separate "tyrosine shortens life" from "the diseases that raise tyrosine shorten life" when the instruments are the disease genes themselves.

A word about the venue, because it explains the three headlines. Aging, sometimes styled Aging-US, is published by Impact Journals of Buffalo, New York. It is indexed in MEDLINE and has a real impact factor, but it was placed on Beall's list of potentially predatory publishers in 2015, its sister journal Oncotarget was dropped from MEDLINE in 2018, and in October 2024 Aging itself received a discontinuation notice from the Web of Science Core Collection, which it is appealing.²² The publisher offers authors a "free post-publication promotion service." That service issued a press release in November 2025 and has evidently kept re-issuing it, which is how a paper from October 2025 became "news" in February, June, and September of 2026. The parental-lifespan GWAS the analysis depends on was also published in Aging.²³ None of this makes the science wrong. It does explain why you keep seeing it.

The Mechanism Contradicts Itself, and the Capsule Isn't in the Data

The paper proposes insulin resistance as the pathway by which tyrosine might shorten life, and this is a reasonable guess, because high blood tyrosine has been a known marker of metabolic trouble since 2011, when Wang and colleagues showed in Framingham that isoleucine, leucine, valine, tyrosine, and phenylalanine predicted future diabetes up to twelve years out.²⁴ But a marker is not a cause, and the same genetic method the lifespan paper relies on has been applied to that exact question with the opposite result.

The more parsimonious reading, which the American Council on Science and Health also arrived at, is that elevated fasting tyrosine is a readout of a liver and metabolic system under strain, not an agent of harm.²⁷ The paper's own dose-response supports this. Below the population mean, higher tyrosine was associated with lower mortality (HR 0.89); only above the mean did it turn harmful (HR 1.10).³ That is the shape you get from a metabolic marker, not from a toxin.

And here is the part that should have been the first line of every article: the study did not measure supplement use. The authors say so. "Our study is not directly related to tyrosine supplement."³ It measured fasting plasma tyrosine, once, in the general British population. The question of whether a capsule could push you into the danger zone is a pharmacokinetic one, and the pharmacokinetics are not on the supplement's side either.

Oral tyrosine produces a spike, not a plateau. At 100 mg/kg, roughly 7 grams, plasma tyrosine roughly doubles and peaks at two hours.²⁸ At 100 to 200 mg/kg in van de Rest's older adults it rose five- to seven-fold, peaked at 90 to 120 minutes, and was declining by 150 minutes.¹⁶ The only chronic-dosing trial I could find, a four-week study of 30 men at 1 to 4 grams per day, found fasting plasma tyrosine "modestly increased" only at the 4-gram dose, with no laboratory or adverse-event signal.²⁹ That trial was designed, funded, and co-authored by Ajinomoto, the world's largest amino-acid manufacturer, so weigh it accordingly. But its finding cuts in an inconvenient direction for the scare: at consumer doses, tyrosine barely moves the fasting compartment that the UK Biobank study measured. You cannot take a 500 mg capsule into the Zhao paper's exposure.

Boring at 2 Grams, Not Boring With the Wrong Prescription

Acute doses up to 150 mg/kg have been tolerated in trials, with the usual complaints being nausea, headache, and gastrointestinal upset, mostly in combination regimens.^16,30 In the stress trials 100 mg/kg tended to lower diastolic blood pressure slightly; one 150 mg/kg study reported raised heart rate and pressure. The Ajinomoto trial set a no-observed-adverse-effect level of 4 grams per day for four weeks.²⁹ Nobody has studied daily use for longer than that. The real risks are interactions.

Two theoretical concerns deserve a sentence each. Tyrosine is the substrate for melanin synthesis, so monographs advise caution in melanoma, but there is no human evidence that supplementation affects the disease in either direction. And a 13-week rat study found liver and kidney changes at 2,000 mg/kg per day, a dose no human takes.³¹ Neither belongs on a warning label. Both belong in an honest accounting of what we do not know.

A Marker, an Emergency Buffer, and Nothing Else Yet