Do nootropic peptides actually cross the blood-brain barrier?

TL;DR

• Passive brain entry needs two things together - molecular weight under roughly 400 daltons and fewer than about eight hydrogen bonds. Most licensed small molecules fail this test.
• Semax is around 800 daltons and Selank around 750 - roughly double the passive-diffusion ceiling before charge is even considered; Cerebrolysin's fragments sit below 10,000 daltons, twenty times that threshold.
• Plasma stability is usually the binding constraint: native tuftsin survives seconds, native angiotensin IV about two minutes, and Semax's own half-life is described in minutes.
• No named, characterised, saturable blood-to-brain transport system exists in the literature for Semax, Selank, dihexa or cerebrolysin.
• Born et al. (2002) is the real human evidence - intranasal melanocortin(4–10), vasopressin and insulin raised CSF concentrations within about 30 minutes with no matching serum rise.
• The best intranasal Semax measurement is a rat study: about 0.09% of administered radioactivity per gram of brain at two minutes, roughly 80% of it intact.
• Kp,uu - the standard currency of CNS drug development - is essentially absent from the published nootropic-peptide literature.

---

Educational content, not medical advice. This article contains no doses, schedules or protocols.

The short answer - and why "it's a small molecule" is the wrong intuition

The most common defence of a nootropic peptide claim is some version of: it's small enough to get in. This is the wrong frame, and it is wrong in an instructive way.

"Small molecule" is a pharmacological category, not a size adjective. In the classical account of passive brain entry - set out most clearly by William Pardridge over several decades of review work - a compound crosses the barrier by lipid-mediated free diffusion only if it satisfies two criteria together: a molecular weight under roughly 400 daltons, and high lipid solubility, which in practice means forming fewer than about eight hydrogen bonds. Most licensed small-molecule drugs fail this test. That is the reason CNS drug development is hard, not an obscure technicality.

Peptides are not in this game at all. A heptapeptide like Semax is around 800 daltons; Selank is around 750. Both are roughly double the passive-diffusion ceiling before charge or hydrogen bonding is even considered. Cerebrolysin is not a single molecule but a mixture, described by its manufacturer as containing free amino acids and peptide fragments below 10,000 daltons - a threshold that circulates in marketing copy as though it were the barrier's cut-off. It is not. It is more than twenty times the passive-diffusion threshold. A 9,000-dalton peptide is not "small enough to cross"; it is a protein fragment.

The barrier also does not sort by intuition. A dipeptide and a thirty-residue peptide are entirely different pharmacokinetic problems, and two peptides of near-identical mass can behave completely differently depending on charge, backbone hydrogen bonding, susceptibility to peptidases, and whether an efflux pump recognises them. Molecular weight is a weak single predictor. Anyone using it as their whole argument is not making an argument.

What the blood-brain barrier actually excludes, and how it decides

The barrier is not a membrane. It is a property of the brain's capillary endothelium, and it runs three filters at once.

The physical filter. Brain capillary endothelial cells are joined by continuous tight junctions, have no fenestrations, and show very low rates of non-specific pinocytosis. In peripheral capillaries, solutes leak between and through cells routinely. In the brain, that route is essentially closed. The endothelium sits within a neurovascular unit - pericytes, astrocyte end-feet, basement membrane - that maintains and reinforces this phenotype.

The metabolic filter. The endothelium is enzymatically active. Peptidases expressed on and within the endothelial cells degrade peptides at the interface itself. A molecule can arrive at the barrier and be dismantled there without ever crossing.

The transport filter. This cuts both ways. Selective carriers move specific nutrients in - glucose via GLUT1, large neutral amino acids via LAT1. Efflux pumps move things out, back into the blood. P-glycoprotein is the best-characterised of these, alongside BCRP and the MRP family. P-gp is the reason many compounds that satisfy Pardridge's size and lipophilicity criteria still fail to accumulate in brain: they get in and are immediately pumped back out. This is well-established human and animal pharmacology, not inference.

One structural nuance matters enormously for reading claims. A few brain regions - the circumventricular organs, including the area postrema and median eminence - have fenestrated capillaries and lack a full barrier. Peripheral signals can act there. Peptides can also act on vagal afferents outside the CNS and change brain activity that way. Which means: a compound can produce a genuine central effect without crossing the blood-brain barrier at all. This is not hypothetical. It is a live and well-documented part of how gut and metabolic peptides signal to the brain, and it is discussed in more detail in GLP-1s and the brain. Hold onto it - it defuses a large share of the reasoning in this space.

Size, charge, lipophilicity - and enzymatic degradation before the barrier is even reached

Three properties govern whether a peptide could passively cross, and a fourth usually decides the question before any of them get a vote.

Charge. At physiological pH most peptides carry ionisable groups. Selank's sequence contains lysine and arginine, both basic; charged species diffuse across lipid membranes poorly. Charge is not a minor modifier here - it is often disqualifying on its own.

Hydrogen bonding. This is the underrated one. The peptide backbone is a hydrogen-bonding machine: every amide linkage contributes both a donor and an acceptor. Each hydrogen bond a molecule forms with water has to be broken to enter a lipid bilayer, and the energetic cost compounds. A seven-residue peptide is carrying far more than the eight-bond ceiling from the backbone alone. This is why peptides fail passive diffusion even when their mass looks tolerable.

Lipophilicity. It can be engineered in. Dihexa is the clearest example in this category: an angiotensin IV analogue deliberately built as a modified dipeptide with a lipophilic acyl group attached, specifically to raise hydrophobicity and reduce peptidase susceptibility. That design intent is documented in the medicinal-chemistry literature. Whether the resulting brain exposure has been characterised in humans is a separate question, addressed further down.

And then, the constraint that usually settles it: the peptide is gone before it arrives. Plasma contains aminopeptidases, carboxypeptidases and endopeptidases that degrade circulating peptides on a timescale of minutes or less. Native tuftsin, the parent of Selank, is reported to survive in plasma for a matter of seconds. Native angiotensin IV, the parent of dihexa, is reported at roughly two minutes. Semax's own plasma half-life is described in the literature as minutes.

This matters more than permeability, and it is the point most often skipped. Brain exposure is not a permeability coefficient - it is a permeability coefficient integrated over the time the compound is actually present in plasma at a given concentration. If the concentration collapses in three minutes, an excellent permeability figure buys very little. Plasma stability is frequently the binding constraint, and the compounds in this category were largely designed around it rather than around permeability. The Pro-Gly-Pro tails on Semax and Selank exist to slow degradation, not to unlock a transport route.

Degradation also creates an interpretive trap. Semax is metabolised to Pro-Gly-Pro, a fragment that is itself biologically active and is reported to persist longer than the parent. If a central effect follows a Semax dose, attributing it to Semax crossing the barrier assumes something that has not been separated out - the effect could belong to a metabolite, and the metabolite is a different molecule with different transport properties. Very few studies in this space distinguish parent from fragment. (Evidence tier: animal pharmacokinetics and mechanistic inference; the metabolite question is generally unresolved rather than answered.)

Active transport: the narrow set of peptides with a known route in

Passive diffusion is not the only way in, and the honest version of this section is that documented routes exist - for a specific and short list of peptides.

Saturable peptide transport systems. Work led by William Banks from the 1980s onward established that certain peptides cross by saturable, specific, stereo-selective transport rather than diffusion. Named systems include one carrying Tyr-MIF-1 and related N-terminal-tyrosine peptides such as the enkephalins (predominantly brain-to-blood), a system handling vasopressin-like peptides, and blood-to-brain transport for peptide T, with LHRH described as bidirectional. These are real, characterised systems with saturation kinetics - which is the diagnostic evidence that a carrier exists rather than passive leak.

Receptor-mediated transcytosis (RMT). Larger ligands can be ferried across in vesicles via receptors expressed on the luminal endothelial surface - the transferrin receptor, insulin receptor and IGF1R are the best-studied. This is the mechanism behind engineered "molecular Trojan horse" biologics. It is worth knowing the ceiling: even purpose-built, heavily optimised transferrin-receptor shuttles are reported in the recent literature to achieve on the order of a few percent of injected dose per gram of brain. That is considered a strong result. It is what deliberate engineering against a validated receptor buys.

Carrier-mediated transport (CMT). GLUT1, LAT1 and similar carriers have tight substrate specificity and handle small nutrients. They are not a general peptide route.

Now the part that has to be said plainly. For the compounds this community actually discusses - Semax, Selank, dihexa, cerebrolysin - there is no named, characterised, saturable blood-to-brain transport system in the peer-reviewed literature. Dihexa's design rationale is explicitly passive: increase lipophilicity, decrease hydrogen bonding, resist degradation. Cerebrolysin's "crosses the barrier because the fragments are under 10 kDa" story is a size argument, and as established above, a size argument at that threshold does not work. Semax and Selank are almost always discussed in terms of the intranasal route precisely because the systemic route is unpromising.

And the guardrail that follows: the existence of a transport route for some peptides is not evidence that any particular peptide uses it. "Peptides can cross via saturable transport" and "this peptide crosses via saturable transport" are separated by an experiment that, for most of these compounds, has not been done.

The intranasal claim - what the olfactory and trigeminal routes can and cannot deliver

This is where the field's claims are loudest and its measurements thinnest, so it deserves the most care.

The mechanism is real. Two anatomical routes connect the nasal cavity to the CNS without passing through the systemic circulation. The olfactory pathway runs from the olfactory epithelium in the olfactory cleft at the roof of the nasal cavity, where olfactory sensory neurons project through the cribriform plate to the olfactory bulb; transport appears to occur largely extracellularly, in the perineural spaces surrounding those neurons, which is why it can happen within minutes rather than at the pace of axonal transport. The trigeminal pathway runs from branches innervating both respiratory and olfactory epithelium into the brainstem at the pons, delivering to more caudal regions. Perivascular transport has also been proposed as a contributor. This is mainstream, published delivery science.

There is genuine human evidence - for a small set of peptides. The landmark study is Born and colleagues in Nature Neuroscience (2002), which administered melanocortin(4–10), vasopressin and insulin intranasally to human volunteers and measured increased cerebrospinal fluid concentrations within about 30 minutes without a corresponding rise in serum - the serum result being the important half, since it argues for direct nose-to-brain transfer rather than systemic absorption followed by barrier crossing. That is a properly designed human study and it should be taken seriously.

It should also be read for what it is. It was three specific peptides. It measured CSF, which is a proxy for brain exposure and not the same thing as brain interstitial fluid or brain tissue. And it does not generalise to other molecules by analogy.

Now the parts the marketing omits.

Deposition geometry. In rodents, olfactory epithelium covers roughly half of the nasal cavity. In humans it covers a small single-digit to low-double-digit percentage, tucked in a cleft at the roof of the nasal cavity that ordinary nasal sprays are poorly designed to reach. Most of what is sprayed lands on anterior respiratory mucosa, is swept posteriorly by mucociliary clearance within minutes, and is swallowed. This single anatomical difference is the largest reason rodent nose-to-brain results overstate what to expect in a human, and it applies to every compound in this category.

The delivered fraction is small even in the favourable species. A Russian pharmacokinetic study of intranasal Semax in rats (Shevchenko and colleagues, Russian Journal of Bioorganic Chemistry, 2006) reported on the order of 0.09% of administered radioactivity per gram of brain tissue at two minutes, with roughly 80% of that radioactivity attributable to intact Semax. Read that carefully: it is a real measurement, it demonstrates that something reaches the brain, and it is a fraction of one percent per gram - in the species with the most favourable nasal anatomy available. (Evidence tier: animal model, radiolabel distribution.)

"Bioavailability" is usually the wrong number. Nasal bioavailability figures quoted for these compounds - and there are some very high ones circulating - generally describe systemic absorption: how much of the dose reached the bloodstream rather than being swallowed. That says nothing about brain delivery. A compound can have excellent nasal bioavailability and negligible CNS exposure. Treat any percentage cited in support of a brain claim as needing its denominator identified.

So what about "it bypasses the blood-brain barrier"? As a statement about anatomy, it is defensible - the olfactory and trigeminal routes do not traverse the brain capillary endothelium. As a claim about a compound, it is incomplete, because bypassing the barrier tells you nothing about how much arrives. Delivery is a quantity. Quantity requires measurement. A mechanism story is not a measurement, and in this field the mechanism story is nearly always offered in place of one.

What a measured CNS concentration would look like, and how rarely anyone measures one

If someone wanted to settle whether a given peptide reaches the human brain, here is what that would take - and why it usually does not exist.

Brain tissue concentration is the gold standard and requires terminal sampling. In practice this means animals only.

Human CSF sampling requires lumbar puncture or an indwelling drain. It is invasive, and lumbar CSF is an imperfect proxy: it is not brain interstitial fluid, gradients exist along the neuraxis, and CSF turnover acts as a sink. It is nonetheless the strongest routinely obtainable human measure, and it is what the Born study used.

Cerebral microdialysis samples brain interstitial fluid directly, but requires a neurosurgically placed probe. It happens in intensive care and epilepsy monitoring populations. It does not happen in healthy-volunteer nootropic research.

PET imaging with a radiolabelled analogue can quantify brain uptake non-invasively, but requires a suitable radioligand, measures total rather than unbound compound, and does not separate parent from metabolite without additional work.

The metric that actually answers the question is Kp,uu - the ratio of unbound drug in brain to unbound drug in plasma at steady state. It is the standard currency of CNS drug development. It is essentially absent from the published literature on nootropic peptides.

What exists instead, for most of this category, is some combination of: rodent radiolabel distribution studies; in vitro barrier models (cell monolayers, artificial membrane assays) that predict permeability without reproducing the intact neurovascular unit; behavioural or cognitive endpoints in animals; and, in the better-documented cases, human clinical or EEG endpoints with no accompanying pharmacokinetics.

Insulin is in that table deliberately. It is not a nootropic peptide and nothing here maps it to a use - it is there because it shows what the evidence bar looks like when it is actually cleared, and how far the rest of the column sits from it.

The honest summary: the compounds with the strongest CNS-penetration evidence in this table are not the ones being marketed as nootropics. For Semax and Selank the case rests on animal work plus human functional endpoints. For dihexa it rests on preclinical work and design rationale. For cerebrolysin it rests on a size heuristic that does not survive scrutiny, with the clinical trial literature - mixed and contested, as covered in Dihexa and Cerebrolysin, honestly assessed - measuring outcomes rather than exposure.

Reading a nootropic-peptide claim: three questions that settle most of them

These are portable. They work on a vendor page, a forum post, a podcast, or a study abstract.

1. Was this measured in humans, or inferred from rodents? Nose-to-brain rodent data does not transfer cleanly to humans, for the anatomical reason set out above. Systemic rodent PK does not transfer cleanly either, because peptidase profiles and barrier transporter expression differ between species. If the only measurement is in an animal, the claim is a hypothesis about humans, and should be stated that way.

2. Was a CNS concentration measured at all - or only a behavioural endpoint? This is the highest-yield question, because "people reported feeling sharper" and "the compound reached the brain" are separate claims and the first does not evidence the second. Peripheral routes to a central effect are real: circumventricular organs lacking a full barrier, vagal afferent signalling, immune and inflammatory modulation, and effects produced by metabolites rather than the parent molecule. If a study measured an outcome but never measured an exposure, it has not addressed penetration - whatever it found.

**3. Does the claimed route have a documented transport mechanism for this compound - or is it asserted?** Ask what specifically is being claimed. A named saturable transport system with published kinetics is a strong claim. Receptor-mediated transcytosis with a named receptor is a strong claim. "It's small enough" is a size heuristic. "It bypasses the barrier intranasally" is a statement about anatomy that leaves the delivered fraction unspecified. "Peptides can cross the barrier" is true of some peptides and tells you nothing about the one in front of you.

Where this deliberately stops. Applying these three questions does not produce a ranking, and it is not meant to. Nothing above says any compound works, that better penetration evidence would mean better effects, or that a reader should prefer one thing over another. Penetration is a necessary condition for a direct central mechanism, not a sufficient one for a benefit - a molecule can reach the brain and do nothing useful there. The output of this article is a sharper filter for claims, not a selection. Which compounds are even lawfully available is a separate matter again, covered in Is Semax legal in 2026?.

Frequently asked questions

If a peptide's plasma half-life is only a few minutes, does that mean it can't reach the brain? Not automatically, but it raises the bar considerably. Brain exposure depends on how permeable a compound is and how long it persists at a given plasma concentration; a very short half-life means the window for transfer is correspondingly narrow. It is also why the intranasal route dominates discussion of these compounds - it is an attempt to route around the plasma problem rather than solve it. But routing around it does not remove the need to measure what arrives. Short half-life plus no CNS concentration data is not evidence of failure; it is an absence of evidence in either direction, and it should be described that way rather than filled in with a mechanism story.

Is "it bypasses the blood-brain barrier" true for intranasal administration? Partly, and the imprecision matters. The olfactory and trigeminal pathways genuinely do not cross the brain capillary endothelium, so as a description of anatomy the statement holds. What it omits is quantity. Human olfactory epithelium occupies a small fraction of the nasal cavity in a cleft that conventional delivery reaches poorly, mucociliary clearance removes much of what is deposited elsewhere, and the fraction reaching brain tissue in the best-studied animal case was well under one percent per gram. "Bypasses the barrier" describes a road. It says nothing about how much traffic arrives.

If someone feels a clear effect, doesn't that prove the compound got into their brain? No, and this is the most common inferential error in the category. A subjective or measurable central effect is consistent with barrier crossing, but it is equally consistent with several other explanations: action at circumventricular organs that lack a full barrier, signalling via vagal afferents, peripheral immune or inflammatory modulation that alters brain state indirectly, activity of a metabolite rather than the administered molecule, and - since these are almost always uncontrolled personal observations - expectancy. Distinguishing between these requires measurement, not introspection.

Does molecular weight tell you anything useful at all? It tells you something, but far less than it is asked to. Under about 400 daltons combined with high lipid solubility is a meaningful passive-diffusion criterion; above it, passive diffusion is unlikely, which is genuine information. What weight cannot do is act as a single sorting variable. Charge, hydrogen-bond count, peptidase susceptibility, plasma protein binding and efflux-pump recognition all move the answer, sometimes decisively. Two peptides of the same mass can behave completely differently. Any claim resting on weight alone - in either direction - is under-specified.

Related reading

• Semax and Selank: the nootropic peptides, honestly - what the clinical history for the two best-documented compounds actually contains
• Dihexa and Cerebrolysin, honestly assessed - the advanced end, where the evidence thins and the claims escalate
• GLP-1s and the brain - a worked example of central effects with contested penetration, including the peripheral-signalling routes described above
• Is Semax legal in 2026? - the regulatory picture after the July FDA advisory committee vote
• Peptides and mood: what the evidence actually says - the delivery problem described above, applied to the mood claims that depend on it

Part of The Complete Guide to Nootropic Peptides.

Educational content, not medical advice. This article describes mechanisms and evidence quality only. It contains no doses, schedules, protocols or administration techniques, it does not recommend or compare compounds for any purpose, and nothing in it should be read as suggesting that any compound is safe, effective or appropriate for any person. The compounds discussed are investigational; none is an approved cognitive treatment, and any use is a matter for a qualified clinician who knows your history. Claims for the Malaysian market are subject to MAB requirements.

Explore it with verified supply. Every Vivre Labs lot is independently tested and the certificate of analysis is published, so you know exactly what is in the vial. Explore Semax.

References

• Pardridge WM. "Drug Transport across the Blood-Brain Barrier." Journal of Cerebral Blood Flow & Metabolism, 2012. (Dual criteria for lipid-mediated free diffusion: molecular weight under ~400 Da and fewer than ~8 hydrogen bonds; role of P-glycoprotein efflux in defeating otherwise-compliant molecules.)
• Banks WA, Kastin AJ. "Saturable transport of peptides across the blood-brain barrier." Life Sciences, 1987; and subsequent Banks reviews on peptides and the blood-brain barrier. (Characterisation of specific, saturable, stereo-selective peptide transport systems, including the Tyr-MIF-1/enkephalin system, vasopressin-like peptide transport, peptide T, and bidirectional LHRH transport.)
• Born J, Lange T, Kern W, McGregor GP, Bickel U, Fehm HL. "Sniffing neuropeptides: a transnasal approach to the human brain." Nature Neuroscience, 2002. (Intranasal melanocortin(4–10), vasopressin and insulin produced increases in human CSF concentrations within approximately 30 minutes without corresponding serum increases.)
• Shevchenko KV, et al. "Kinetics of semax penetration into the brain and blood of rats after its intranasal administration." Russian Journal of Bioorganic Chemistry, 2006. (Radiolabel distribution study; on the order of 0.09% of administered radioactivity per gram of brain at 2 minutes, with approximately 80% attributable to intact Semax.)
• Reviews of intranasal nose-to-brain delivery mechanisms covering olfactory and trigeminal perineural transport, perivascular contribution, and the anatomical differences in olfactory epithelium coverage between rodents (approximately half the nasal cavity) and humans (a small single- to low-double-digit percentage), including differential targeting of olfactory versus respiratory epithelium. Expert Opinion on Drug Delivery, 2025, and related delivery-science literature.
• Reviews of receptor-mediated transcytosis for brain delivery of therapeutics, covering transferrin receptor, insulin receptor and IGF1R shuttles and reported delivered fractions on the order of a few percent of injected dose per gram of brain for optimised constructs. Frontiers in Drug Delivery, 2024, and related literature.
• Medicinal-chemistry literature on metabolically stabilised angiotensin IV analogues (dihexa/PNB-0408), describing the design rationale of increased hydrophobicity, reduced hydrogen bonding and resistance to peptidase degradation, and the very short plasma half-life of the parent angiotensin IV.
• Manufacturer and review descriptions of Cerebrolysin as an enzymatically derived porcine brain protein hydrolysate comprising free amino acids and peptide fragments below 10,000 daltons.