Semax: a Russian stroke drug sold as a focus spray, and what its numbers actually show

A stroke drug from Moscow, sold in America as a focus spray

Semax was built at the Institute of Molecular Genetics in Moscow from a fragment of ACTH, the hormone your pituitary releases under stress. ACTH has two jobs: it tells the adrenal gland to make cortisol, and it acts directly on brain tissue. The Moscow group cut the molecule down to the part that acts on the brain and threw away the part that drives cortisol. What is left is four amino acids of ACTH — positions 4 to 7 — with three more stuck on the end.

In Russia it is on the list of essential medicines and it is given for stroke, brain injury, optic nerve damage and, in children, attention problems. In the United States it has never been submitted to the FDA for review, so it is not an approved drug for anything. The Wikipedia entry and the vendor pages agree on this history; the disagreement is entirely about what the history proves.

That gap between forty years of clinical use in one country and near-zero regulatory footprint in another is the single most important fact about Semax, and it cuts both ways. Russian hospital use is real evidence that thousands of people have taken it without an obvious safety disaster. It is not evidence that it works, because most of that use was never tested against a placebo in a trial anyone outside Russia can read.

Seven amino acids, and the last three exist only to stop your body eating it

The sequence is Met-Glu-His-Phe-Pro-Gly-Pro. The first four letters are the piece of ACTH that acts on brain tissue. The last three — proline, glycine, proline — do nothing on their own. They are there because a bare four-amino-acid peptide is chewed up by enzymes in the blood within seconds. The tail slows that down.

This matters for two practical reasons. First, it explains why the nose route dominates: sprayed into the nose, part of the dose reaches brain tissue along the olfactory route without having to survive a trip through the liver. Second, it explains why an even more protected version exists. N-acetyl Semax amidate — sold as NA-Semax — has a chemical cap on each end, lasts longer, and is a different product with a different dose response. People who report switching to it consistently report a longer effect, and one person reported their thinking got worse on it. Treating the two as interchangeable is the most common mistake in the self-reported record.

The signal it pulls: one growth factor, and the receptor it docks into

The mechanism claim rests on one growth factor. Brain-derived neurotrophic factor, or BDNF, is the protein nerve cells use to grow new connections, keep existing ones alive, and rebuild after damage. It works by docking into a receptor on the cell surface called TrkB. When BDNF binds TrkB, the receptor switches on and starts a chain of signals inside the cell that ends in new protein being built at the synapse.

Injured nerve tissue runs short of this signal. After a stroke, a concussion or a crushed nerve, BDNF levels in the affected region drop, and the cells that survive have fewer resources to rebuild connections. Chronic stress does the same thing more slowly: in rats under weeks of unpredictable stress, hippocampal BDNF falls, adrenal glands enlarge, and the animals stop seeking things they used to like.

Semax raises that signal. That is the whole mechanistic case, and the numbers behind it are specific.

One rat spray, four numbers, and the whole mechanism case

The 2006 study by Dolotov and colleagues is the load-bearing measurement. A single dose of Semax sprayed into a rat's nose, 50 micrograms per kilogram of body weight, produced in the hippocampus:

• BDNF protein up 1.4-fold
• TrkB receptor switched on — phosphorylation up 1.6-fold
• BDNF messenger RNA (exon III) up 3-fold
• TrkB messenger RNA up 2-fold

Four numbers from one dose in one brain region. The same group found BDNF protein rising in the basal forebrain too. Every claim you will read about Semax and neuroplasticity traces back through this measurement or a close relative of it.

Scale it honestly. A 1.4-fold rise in a protein is a real biological effect and a modest one. It is not a switch being thrown; it is a dial being turned. Whether turning that dial by 40 per cent for a few hours changes how a person thinks or heals is exactly the question the human studies were too small to answer.

Other animal work fills in the picture around it. A 2025 rat study found Semax changes which genes for neurotrophic factors are switched on, and that it stirs the dopamine and serotonin systems. A 2017 rat study looking across all biological processes found the immune response was the process most changed by the peptide, which nobody has followed up properly in people. A 2024 study in rats under chronic unpredictable stress found Semax reversed or blunted three separate consequences of that stress: the loss of interest in rewards, the enlargement of the adrenal glands, and the fall in hippocampal BDNF. In a mouse model of Alzheimer's disease, Semax and a derivative improved performance on three separate tests — open field, novel object recognition, and the Barnes maze.

There is also a chemistry finding that has nothing to do with BDNF. Semax grabs copper ions, and in artificial membrane models it interferes with the way copper-bound amyloid beta assembles into fibres. That is a test-tube result about a molecule, not a result about a brain.

The counter-example is worth as much as the wins. A 2025 review of Semax in Parkinson's models found the results inconsistent across different rat models, found that low doses did not improve movement problems, and noted that no clinical trial has ever tested it in people with Parkinson's.

The mouse with a crushed spinal cord is the closest thing to a nerve-injury result

A 2025 paper in the British Journal of Pharmacology took female mice with spinal cord injuries and gave them Semax. Recovery of function improved. The mechanism the authors traced was not BDNF. It ran through the gene for the mu opioid receptor, Oprm1: Semax kept a protein in that pathway from being tagged for destruction, which in turn shut down a form of inflammatory cell death — cells rupturing and spilling their contents, which recruits more inflammation and kills the neighbours.

For anyone reading this because of a damaged nerve rather than a foggy morning, this is the study that matters most, and its limits need saying in the same breath. It is mice. It is one paper. The injury was a surgical crush to the cord under laboratory conditions. And the pathway it found is not the pathway the marketing talks about.

What has actually been measured in people: 24 healthy volunteers and 110 stroke patients

Two human studies carry the weight.

The brain-imaging pilot. Twenty-four healthy people were given a 1 per cent Semax solution into the nose, 1.2 milligrams in total, or a placebo. Resting brain scans showed a larger signal in the front-midline part of the default mode network — the set of regions that stay active when you are not doing a task — in the Semax group. A related study found the volume of that same front region larger in the Semax group. This is a real, placebo-controlled measurement in people. It measures a brain signal, not a thought, and nothing about it tells you whether a person performed better.

The stroke study. One hundred and ten patients recovering from an ischaemic stroke were given Semax in two ten-day courses at 6,000 micrograms a day with a twenty-day gap between them. Plasma BDNF rose and stayed up for the length of the study, regardless of whether rehabilitation had started early or late. Patients with higher BDNF had better movement scores and better Barthel index scores, which is the standard measure of how much of daily life a person can manage alone.

Read that last sentence carefully, because the sentence that gets written about it in marketing copy is a different sentence. The study measured a protein in the blood and found it correlated with recovery. It did not show that Semax caused the recovery. The peptide raised BDNF; better BDNF went with better outcomes; the causal step between them was not tested.

Eleven people, tired, doing an attention test

The third human data point is the smallest and the most honest about itself. Eleven healthy participants were tested on selective attention and short-term memory, with the effect showing up mainly under fatigue. Eleven people is a pilot. It is quoted widely as evidence Semax sharpens cognition in healthy adults; what it supports is that somebody should run the study properly.

One physician who reviewed the whole set publicly graded it: D+ for the focus and nootropic claims, C- for the stroke and blood-vessel evidence. That grade is one person's judgement, not a measurement, and it lands close to where the numbers above sit.

Forty-three people wrote down what happened, and the split is wide

Sixty-eight self-reported sources were collected for this page from Reddit, X, YouTube and Instagram. Twenty-five of them are explainers, vendor posts, dosing questions, guides, news items or general commentary that report no personal outcome, so they are not counted here. Forty-three describe what happened to a specific person.

Of those 43: 22 reported a benefit, 10 reported a benefit that came with a cost or faded, 4 reported nothing at all, and 7 reported harm as their main experience.

This is a self-reported record. Nobody was blinded, nobody was randomised, the people who felt nothing are less likely to write a post than the people who felt transformed, and several accounts describe taking Semax alongside Selank, stimulants or other peptides, which makes the cause of any single report unclear. It is worth counting anyway, because it is the only large body of information about what happens when this compound is taken by people outside a hospital, and because the split it shows — roughly half clear benefit, a quarter benefit-then-problem, a sixth harm — is not the split the sales pages describe.

The 22 who said it worked

The strongest reports are about mental performance under load, not mood. One person doing eight weeks on, eight weeks off wrote that at work "I perform so much better at high level mental tasks... my recall and articulation are noticeably improved." Another, after finishing a bottle of the 0.1 per cent spray, described fixing long-standing focus and memory problems and feeling, months later, "like I upgraded."

Speed of onset comes up repeatedly. One person on X described visual sharpness and mental clarity arriving "within just a few minutes," lasting from 8:30 in the morning until about 2 in the afternoon, with no crash. Another reported the room going quieter and "less friction between my thoughts and my work," again with no spike and no crash. One person taking 250 micrograms noticed within about thirty minutes that they were starting tasks that had been sitting for weeks.

Two reports are about damage rather than performance. A person in a traumatic brain injury forum wrote that after a couple of weeks on the nose spray they "did get some memory back." A person with epilepsy wrote that they remember more from the days they take it, focus better, and noticed their speech improving.

Two reports are about dose-finding rather than the drug. One person moved from 200 to 500 to 700 micrograms a day under the skin, found the effect only arrived at the higher end, and has held at 700 for two months. Another compared routes directly: the nose was faster but weaker, the needle took 30 to 60 minutes and was "WAY more consistent."

The 10 who got a result and then lost it

This group is the most informative and the least quoted. The pattern is a strong first response followed by tolerance, a flat mood, or a cost that outweighed the benefit.

Tolerance is the commonest version. One person: "the effect only lasts about 2 weeks, 3 at best before you have to cycle it off." Another, injecting two peptides daily, described "amazing clean focused energy" that faded so fast that at 2 milligrams of each "it was doing nothing," and stopped. A third ran 600 micrograms of nose spray daily for two to three months, called the drive and curiosity it produced remarkable, and in the same post described feeling detached from their surroundings.

Emotional flattening is the second version. One person on about 1 milligram under the skin for a few months got real relief from focus problems, energy and fatigue-driven low mood, then "started to get emotionally blunted" badly enough that it affected their relationship. Another reported hair loss alongside an effect they otherwise liked.

The short-window version: one person felt their senses sharpen for fifteen minutes and then crash into tiredness. Another described strong focus that was "useless past early afternoon or I can't sleep."

The 4 who felt nothing

Four accounts report no effect. "I literally felt zero effects." "I just didn't get the adderall-like effects." One person tried up to 1 milligram and rated it 6 out of 10, saying it helps "a little." One tried it daily, found the side effects acceptable, and stopped because something else worked better.

The 7 who got worse, and the pattern in what went wrong

Seven reports are primarily about harm, and they cluster into four kinds.

Headache. Two people: one woke after a first-ever dose "extremely groggy and tired" with a headache most of the day; one wrote that the only thing they get from Semax is a migraine.

Crashing fatigue. One person, one spray at 6 a.m., described fatigue arriving around noon "as if I drank all night and then woke up with no sleep," and called it intense.

Low mood. One person taking 200 micrograms under the skin reported getting moody and feeling down on it.

Hair loss and worse thinking. One person reported two 30-day runs followed by a year of hair loss. One reported that on the N-acetyl amidate version "the quality of my cognition has declined since starting."

One further report describes passing out after moving from 250 to 500 micrograms, with the poster noting the higher dose had otherwise felt fine. A single fainting episode in a self-reported record is not a rate. It is the kind of event that would be caught in a trial and has not been.

The theme worth extracting: the reported harms are nearly all central nervous system effects on the same axis as the reported benefits — arousal, mood, sleep — which is what you would expect from something that acts on brain signalling rather than an off-target effect somewhere else in the body. The two hair-loss reports are the exception and have no counterpart anywhere in the published literature.

What the tolerance reports mean if you are planning to take it for months

Tolerance appears in the self-reported record and nowhere in the published studies, because the published studies are short. The Russian stroke schedule is instructive on its own terms: two ten-day courses with a twenty-day gap. That is a cycled protocol designed by people giving the drug in hospitals, and it looks nothing like the daily indefinite use most self-reported accounts describe. Whether the gap is there because of tolerance, cost, or hospital practice is not documented in the material available here.

Nose spray against a needle, and the doses people actually use

Published human doses: 1.2 milligrams total, into the nose, in the imaging pilot. 6,000 micrograms a day for ten days in the stroke study. That upper figure is five times what almost anyone in the self-reported record takes.

The doses people report using: 200 to 800 micrograms a day into the nose, or 250 micrograms to 1 milligram a day under the skin. Two strengths of nose spray circulate, 0.1 per cent and 1 per cent, and confusing them is a ten-fold error. One person in the record misread two bottles and took far more than intended.

Route changes the shape of the response, consistently across accounts. Nose: faster, weaker, shorter. Needle: slower to arrive, steadier, longer. That pattern is what you would expect from where each route delivers the peptide, and it is reported often enough to be worth planning around.

Where it stands with the FDA after the April 2026 list change

Semax is not an approved drug in the United States and has never been submitted for approval.

The compounding position changed in 2026 and the change is widely misdescribed. The FDA's revised 503A bulk substances document, published 15 April 2026, removed Semax — both the free base and the acetate salt — from Category 2, the list of substances flagged with significant safety concerns. Removal from Category 2 is not permission. It moved the substance into a review process: the Pharmacy Compounding Advisory Committee took up Semax-related bulk drug substances at its meeting on 24 July 2026, alongside Epitalon and several others.

So the accurate statement is that Semax sits between lists. It is not on the approved bulks list, and it is no longer flagged in Category 2. Any page telling you the FDA banned it, and any page telling you the FDA cleared it, is describing a document that does not exist.

What a serious test would look like, and why nobody has run one

The study that would settle the focus question is not exotic: two hundred healthy adults, randomised, placebo-controlled, one dose into the nose, a standard attention battery, and a measurement of the same brain signal the 24-person pilot found. The study that would settle the peripheral-nerve question is harder and more valuable, because every measurement in this literature was taken in brain tissue and none in a peripheral nerve: an animal model of peripheral nerve injury, with nerve conduction and recovery measured against untreated controls.

Neither has been run, and the reason is structural rather than scientific. The molecule is decades old and unpatentable in its original form, the Russian clinical record does not translate into a US filing, and the American market for it works fine without a trial. That is a commercial fact about the compound, not a fact about whether it works — and it is the reason the honest version of this page has to stop where it does.

Every source behind the numbers above, and how to pull them

Thirty-four published sources are attached to this page: 13 indexed studies, 7 reviews, 12 clinical or regulatory pages, 1 news item, and the encyclopaedia entry. Sixty-eight self-reported sources are attached, of which 43 are counted above with the platform, the quote and the date on each card. Every number in this page traces to one of them, and the cards are on this page rather than in a bibliography so that a claim and the thing it rests on can be read together.