Energy, Cognition & Mood
Best Peptides for Neuroprotection & Brain Health: Evidence (2026)
A clinical, evidence-graded ranking of the peptides marketed for neuroprotection — Cerebrolysin, Semax, ARA-290, SS-31/elamipretide and P021 — separating a modest, mixed human signal from elegant but preclinical-only mechanism.
CerebrolysinSemaxARA-290SS-31evidence-graded
The quick verdict
An honest, evidence-graded ranking of the peptides marketed for neuroprotection — and why none is a proven way to durably protect or restore the brain.
- Best overall
- Cerebrolysin — The only peptide here with a large, genuine human RCT/meta base and a real early-recovery signal in acute stroke — though its primary endpoints were neutral, functional-independence data are inconsistent, and its dementia evidence is 'very low quality' (Grade B).
- Best value
- Semax — The best-evidenced synthetic peptide for the cost, with decades of Russian acute-stroke use and a 110-patient BDNF/recovery trial; the preclinical-only SS-31 and P021 have no human neuroprotection data, so they justify no premium.
- Best for Peripheral small-fiber neuropathy (a nerve-repair use, NOT central brain protection; clinician-supervised, not self-use)
- ARA-290 (cibinetide) — It has the cleanest randomized human data of the group — but for peripheral small-fiber neuropathy, where it promoted objective nerve-fiber regeneration; it has no human brain-protection trial, so this is a nerve use, not a neuroprotection endorsement.
How we evaluated
We ranked each peptide strictly by the strength and volume of human evidence for a central neuroprotective endpoint — stroke recovery, TBI recovery, or slowing neurodegeneration — not fame, mechanism elegance, or evidence for unrelated conditions. Human trials outrank animal data, which outrank in-vitro and mechanism-only reasoning; anecdote and vendor copy do not count. Because no peptide here has Grade-A human neuroprotection data, we grade each by its best human data touching the brain and flag where that signal is neutral, single-country, peripheral rather than central, or absent. Evidence grades follow PeptideVox's A-to-D ramp (A = human RCT/meta; B = lower-tier human; C = preclinical only; D = anecdotal/marketing).
- Human neuroprotection evidence. Whether any published human study used a central neuroprotective endpoint (stroke/TBI recovery, cognition in neurodegeneration), in what population, and with what result — including neutral primary endpoints.
- Evidence quality & independence. Trial size, blinding, controls, region/language diversity, replication, GRADE certainty, and whether positive results correlate with manufacturer sponsorship.
- Mechanistic coherence. Whether a plausible mechanism (neurotrophic mimicry, innate-repair/EPO signaling, mitochondrial/cardiolipin stabilization, anti-apoptosis) supports the claim, ideally with a human correlate.
- Safety & fit. Tolerability, route, serious-adverse-event signals, immunogenicity/product-integrity risk, and uncharacterized long-term data.
- Regulatory honesty. FDA approval/compounding status and WADA exposure, stated plainly rather than glossed over.
Rating scale: 1-5 stars reflecting human-evidence strength for central neuroprotection specifically: 5 = robust, independent human RCT data with a consistent functional benefit; 1 = mechanism or marketing only. Star ratings track the per-item evidence grade and its trustworthiness.
Last verified .
At a glance
| # | Name | Evidence | Rating | Best for | Pricing |
|---|---|---|---|---|---|
| 1 | Cerebrolysin | B | 2.5 | Understanding the ceiling of peptide evidence for neuroprotection — a real, extensively studied adjunct for injured brains abroad, not a proven neuroprotectant | No US NDA; legal here only as a non-human research reagent; not a sourcing recommendation |
| 2 | Semax | B | 2.0 | Those wanting the best-evidenced synthetic peptide for acute-stroke neuroprotection, with eyes open to its single-country, under-replicated evidence base | Research chemical / not FDA-approved; not a sourcing recommendation |
| 3 | ARA-290 (Cibinetide) | C | 2.0 | Understanding why the cleanest human peptide data in this group does not equal brain protection — a peripheral-nerve signal, not a neuroprotectant | Research chemical / not FDA-approved; not a sourcing recommendation |
| 4 | SS-31 (Elamipretide) | C | 1.5 | Illustrating how a mechanistically attractive, FDA-approved molecule can still be preclinical-only for the brain — not for anyone seeking human neuroprotection proof | Research chemical / not FDA-approved for neuroprotection; not a sourcing recommendation |
| 5 | P021 (P021) | C | 1.0 | Understanding why an attractive preclinical neurogenic mechanism, absent any human data, cannot support a neuroprotection claim | Research chemical / not FDA-approved; not a sourcing recommendation |
Cerebrolysin
The only large human RCT base — but neutral primary endpoints and mixed function
Cerebrolysin is the most clinically studied peptide in this category by a wide margin, which is why it ranks first even though its verdict is best summarized as "studied extensively, proven inconclusively." It is not a single molecule but a standardized mixture of low-molecular-weight peptides (roughly 25%) and free amino acids (roughly 75%) produced by enzymatic hydrolysis of purified porcine brain, and it is given parenterally only by slow IV infusion or intramuscular injection. It is approved in 50-plus countries (Austria, Russia, China, South Korea and others) but is not FDA-approved. The human picture is genuinely mixed. In acute ischemic stroke, the largest single trial, CASTA (n=1,070, 30 mL IV daily for 10 days), was neutral on its primary combined global endpoint, and a 2025 meta-analysis of 14 RCTs (n=2,884) found a small, statistically significant early neurological improvement (NIHSS +1.39 points; GRADE moderate) that did not translate into significant functional independence (mRS 0-2). An earlier 7-RCT meta-analysis (n=1,779) likewise failed to show superiority on functional scales. In vascular dementia a Cochrane review of six RCTs found benefit on combined cognition and global function but rated all outcomes 'very low quality' with high bias and industry funding, and in mild-to-moderate Alzheimer's a meta-analysis found a four-week cognitive benefit that faded by six months. The CAPTAIN TBI meta-analysis (n=185) reported a small-to-medium benefit on a multidimensional outcome ensemble. Its safety is its strongest suit — no excess serious adverse events or mortality and a reduced risk of hemorrhagic transformation. But the primary endpoints of its largest trials were neutral, functional outcomes are inconsistent, the dementia evidence is 'very low quality,' and most positive trials are industry-funded — which is exactly why it earns B, not A.
Strengths
- The only peptide here with a large, genuine human RCT/meta base (dozens of trials, several meta-analyses across stroke, TBI, dementia)
- Small but statistically significant early neurological-recovery signal in acute stroke (2025 meta-analysis, NIHSS +1.39 points, GRADE moderate)
- Benign safety record — no excess serious adverse events or mortality and a reduced risk of hemorrhagic transformation (GRADE high certainty)
- Coherent multimodal neurotrophic mechanism (neurogenesis, angiogenesis, anti-apoptosis, anti-inflammation)
Weaknesses
- Neutral primary endpoint in its largest stroke RCT (CASTA, n=1,070), and no significant functional-independence (mRS 0-2) benefit in the 2025 meta-analysis
- Cochrane rated its vascular-dementia evidence 'very low quality' with high bias and industry funding; the Alzheimer's cognitive benefit faded by six months
- Parenteral only (IV/IM), porcine-derived (anaphylaxis/theoretical prion concern), not FDA-approved and with no US compounding pathway
- Best for
- Understanding the ceiling of peptide evidence for neuroprotection — a real, extensively studied adjunct for injured brains abroad, not a proven neuroprotectant
- Pricing
- No US NDA; legal here only as a non-human research reagent; not a sourcing recommendation
Source: Patel et al., Cureus 2025 meta-analysis (PMC12465088)
Semax
Decades of Russian acute-stroke use — but no Western RCT
Semax is a Russian-developed synthetic heptapeptide (Met-Glu-His-Phe-Pro-Gly-Pro), an ACTH(4-10) analog engineered to retain neurotropic activity while shedding the hormonal effects of the parent fragment. It has been a registered drug in Russia since 1994 for ischemic stroke, transient ischemic attack, cognitive disorders and optic-nerve disease, making it the second-most clinically used peptide here, but its evidence base is almost entirely Russian-language and under-replicated. Its strongest clinical data are in acute ischemic stroke: the foundational Gusev work (multicenter, roughly 200 patients) reported that intranasal Semax added to standard care improved neurological deficit scores and recovery with reduced infarct evolution, and a PubMed-indexed study of 110 post-stroke patients (two 10-day courses of 6,000 micrograms per day) raised plasma BDNF, speeded functional recovery and improved motor performance, especially with early rehabilitation. A Russian meta-analysis concluded that 1% intranasal Semax in the acute period reduces neurological impairment and improves mobility and independence — while explicitly calling for a multicenter, well-powered, double-blind trial that does not yet exist. The animal mechanism is consistent and well-replicated: Semax upregulates BDNF and NGF across hippocampus, cortex and retina, and in a rat photothrombotic stroke model reduced infarct size and improved learning. It ranks second, and earns B (a B-minus for the brain specifically), because the human signal is real but small, predominantly Russian-language, variably blinded, generated within a few laboratories, and never validated in a Western double-blind RCT; healthy-person 'cognitive enhancement' claims are essentially anecdotal or extrapolated.
Strengths
- Genuine human clinical pedigree in acute ischemic stroke over three decades of Russian use, with a 110-patient trial raising plasma BDNF and accelerating recovery
- Robust, well-replicated animal mechanism via BDNF/NGF induction and post-ischemic neuroprotection
- Non-invasive intranasal route exploiting nose-to-brain transport, with CNS effects outlasting the minutes-long plasma half-life
- Generally well tolerated in the Russian clinical record (occasional mild nasal irritation)
Weaknesses
- Human data are small, predominantly Russian-language, variably blinded, and never validated in a Western double-blind RCT — capped at Grade B-
- No rigorous Western safety dataset; products sold outside Russia are unregulated 'research' material with purity/sterility concerns
- Healthy-person cognitive-enhancement claims are anecdotal or extrapolated (Grade C-D), not proven neuroprotection
- Best for
- Those wanting the best-evidenced synthetic peptide for acute-stroke neuroprotection, with eyes open to its single-country, under-replicated evidence base
- Pricing
- Research chemical / not FDA-approved; not a sourcing recommendation
Source: Gusev et al., Zh Nevrol Psikhiatr 2018 (PMID 29798983)
ARA-290 (Cibinetide)
Cleanest randomized human data — but for peripheral nerves, not the brain
ARA-290 (cibinetide) is an 11-amino-acid peptide modeled on a non-erythropoietic face of erythropoietin, designed to bind the innate repair receptor — a heteromer of the EPO receptor and the beta-common receptor that is upregulated only in injured or inflamed tissue — triggering anti-apoptotic and anti-inflammatory signaling without stimulating red-cell production, and therefore without EPO's thrombotic or hematologic risk. It has the cleanest randomized human data of this entire group, and it earns an honest split grade because that data concerns the wrong organ for this article. In a randomized, double-blind, placebo-controlled pilot of 22 sarcoidosis patients with small-fiber neuropathy (2 mg IV three times weekly for four weeks), ARA-290 improved neuropathy symptom and quality-of-life scores, and subsequent Phase 2 work (4 mg subcutaneously daily for 28 days) in sarcoidosis- and diabetes-associated small-fiber neuropathy reported increased corneal nerve-fiber area and regenerating GAP-43-positive skin fibers — objective evidence of small-nerve-fiber regrowth, though pain did not consistently separate from placebo in dose-ranging. That is genuine Grade-B human evidence, but for peripheral nerves. Preclinically the EPO-derived/innate-repair-receptor mechanism is neuroprotective in CNS injury models, which is the basis for the 'neuroprotection' marketing — but there is no human trial of ARA-290 for stroke, TBI or any central neurodegenerative disease, so for the brain specifically it is Grade C. It never reached Phase 3 or approval; Araim Pharmaceuticals held orphan designation for sarcoidosis small-fiber neuropathy, but the program stalled after roughly 2020. The honest read: do not conflate nerve repair in the skin or cornea with protecting the brain.
Strengths
- The cleanest randomized, double-blind, placebo-controlled human data of the group — including objective markers of small-nerve-fiber regrowth
- Mechanistically elegant: targets the innate repair receptor upregulated only in injured tissue, so signaling is injury-restricted
- Notably benign safety and, crucially, does not raise hemoglobin or hematocrit — sidestepping EPO's thrombotic risk
Weaknesses
- Its human evidence is for peripheral small-fiber neuropathy, not central neuroprotection — Grade C for the brain
- Zero human trials for stroke, TBI or any central neurodegenerative disease; central claims rest on animal models only
- Never reached Phase 3 or approval; orphan-designation-only, developmentally stalled, sold as 'research use only'
- Best for
- Understanding why the cleanest human peptide data in this group does not equal brain protection — a peripheral-nerve signal, not a neuroprotectant
- Pricing
- Research chemical / not FDA-approved; not a sourcing recommendation
SS-31 (Elamipretide)
Compelling preclinical TBI data — but no human brain trial
SS-31 (elamipretide) is a mitochondria-targeted tetrapeptide that crosses membranes, concentrates in mitochondria and binds cardiolipin to stabilize cristae, support electron-transport efficiency, lower reactive-oxygen-species and inhibit the mitochondrial permeability transition pore — an organelle-level, root-cause mechanism highly relevant to the energy-starved, oxidatively stressed neuron. It is the most pharmaceutically mature peptide on this list overall, the first mitochondria-targeted peptide to win FDA approval (marketed as Forzinity, September 2025) — but for neuroprotection its human evidence is essentially nonexistent, which is why it ranks fourth despite its maturity. There is no completed human RCT of elamipretide for stroke, TBI, dementia or any neurodegenerative cognitive endpoint. Its human record is bioenergetic and rare-disease: a single infusion acutely raised muscle ATP production about 27% in older adults (an effect gone by day 7, with no fatigue benefit), and across four large RCTs in mitochondrial myopathy, heart failure, acute MI and dry AMD it missed every primary endpoint, with its lone approval resting on a 12-patient Barth-syndrome program; a Phase 3 program in Friedreich's ataxia is ongoing but unreported. The neuro case is entirely preclinical and, in fairness, strong: in a mouse TBI model SS-31 reversed mitochondrial dysfunction, reduced ROS, restored superoxide dismutase, lowered cytochrome-c release and DNA damage, attenuated neurological deficits and neuronal apoptosis, and upregulated SIRT1/PGC-1alpha; it is also protective in rodent ischemia and improved learning in sleep-deprived mice. But all of that is rodent or in-vitro. For brain protection it is Grade C, and approval for Barth syndrome does not make it a proven neuroprotectant; 'research-grade SS-31' sold online is not the approved drug.
Strengths
- Elegant, root-cause mitochondrial mechanism (cardiolipin stabilization, ROS reduction) highly relevant to injured neurons
- Strong, coherent preclinical neuroprotection package in mouse TBI and rodent ischemia models
- Benign trial safety profile (injection-site reactions, transient eosinophilia); the first mitochondria-targeted peptide to reach FDA approval
Weaknesses
- No completed human RCT for stroke, TBI, dementia or any neurodegenerative cognitive endpoint — brain neuroprotection is preclinical only (Grade C)
- Missed the primary endpoint in four large non-neuro RCTs; its approval is for Barth syndrome only, resting on a 12-patient program
- No neuro-specific human safety dataset; 'research-grade SS-31' is unapproved, non-pharmaceutical material
- Best for
- Illustrating how a mechanistically attractive, FDA-approved molecule can still be preclinical-only for the brain — not for anyone seeking human neuroprotection proof
- Pricing
- Research chemical / not FDA-approved for neuroprotection; not a sourcing recommendation
P021 (P021)
Preclinical only — single-lab rodent data, zero human trials
P021 (P021) is a synthetic CNTF-derived tetrapeptide (Ac-DGGLAG-NH2) engineered by Khalid Iqbal's group by epitope-mapping the active region of ciliary neurotrophic factor, then adamantylating it for blood-brain-barrier penetration and protease resistance; it is orally bioavailable. It ranks last because its entire evidence base is preclinical rodent work from a single laboratory. There has never been a Phase 1/2 or RCT of P021 in humans — no human efficacy, pharmacokinetic or safety data exist. In triple-transgenic (3xTg-AD) Alzheimer's mice, chronic oral P021 — given preventively from 3 months or therapeutically from 12 months — prevented or rescued cognitive deficits on the Morris water maze, reduced tau hyperphosphorylation, and sustained hippocampal neurogenesis and dendritic/synaptic complexity, largely independent of amyloid-plaque load, with benefits also reported in TBI and age-related cognitive-decline models. Mechanistically it stimulates hippocampal neurogenesis, induces BDNF-TrkB signaling and attenuates tau hyperphosphorylation via the BDNF-TrkB-PI3K-AKT-GSK-3beta axis — without the antibody response and systemic toxicity that doomed full-length CNTF as a drug. That is an internally consistent and biologically elegant package, but a strong single-lab rodent signal is the starting line, not the finish. On dosing, only rodent figures exist (chronic oral/dietary administration over months); no human dose, route or schedule has ever been established. On safety there is no human data; preclinically it avoided the antibody response and systemic toxicity of full-length CNTF, but unregulated 'research' P021 carries the standard impurity and contamination concerns. Any marketed 'P021 brain-regeneration' claim is extrapolation far ahead of an evidence base that contains no human data whatsoever.
Strengths
- Orally bioavailable and blood-brain-barrier-permeable in animal models — a practical delivery profile in principle
- Broad, coherent preclinical package: stimulated neurogenesis, induced BDNF-TrkB signaling, lowered tau hyperphosphorylation and improved maze learning across rodent Alzheimer's, TBI and aging models
- Engineered to avoid the antibody response and systemic toxicity that doomed full-length CNTF (single-lab, animal-level reassurance)
Weaknesses
- Zero human trials of any phase — no human efficacy, pharmacokinetic or safety data exist; the weakest human evidence base here
- Entire efficacy record is preclinical and largely single-lab (the originating Iqbal laboratory)
- No established human dose, route or schedule; human safety entirely unknown, with unregulated 'research' material carrying impurity concerns
- Best for
- Understanding why an attractive preclinical neurogenic mechanism, absent any human data, cannot support a neuroprotection claim
- Pricing
- Research chemical / not FDA-approved; not a sourcing recommendation
Frequently asked
Is any peptide actually proven to protect the brain in humans?
Not at the high-confidence level of a randomized trial or meta-analysis showing a consistent functional benefit. Cerebrolysin has the largest human dataset of any peptide here and a benign safety profile, but its biggest stroke RCT (CASTA, n=1,070) was neutral on the primary endpoint, its functional-outcome data are inconsistent, and Cochrane rated its vascular-dementia evidence 'very low quality.' The other candidates are weaker still: Semax's human data are almost entirely under-replicated Russian trials, ARA-290's clean human data are for peripheral nerves rather than the brain, and SS-31 and P021 have no human neuroprotection trials at all. From a root-cause standpoint, the best-evidenced neuroprotection remains vascular-risk control, sleep, exercise and diet — not an injectable peptide.
Which peptide has the best evidence for stroke recovery specifically?
Cerebrolysin, by evidence volume — it has the most randomized trials and a 2025 meta-analysis of 14 RCTs (n=2,884) showing a small but statistically significant early neurological improvement on the NIHSS. Crucially, that early signal did not translate into significantly better functional independence (mRS 0-2), and an earlier 7-RCT meta-analysis found no superiority on functional scales either. Its safety record is its strongest suit, with no excess serious adverse events or mortality and a reduced risk of hemorrhagic transformation. Semax has decades of Russian clinical use in acute ischemic stroke, but no Western double-blind RCT exists. So Cerebrolysin is best described as extensively studied with a modest early-recovery signal, not a proven functional cure.
SS-31/elamipretide is FDA-approved now — can I use it for brain health?
No. Its September 2025 approval (marketed as Forzinity) is for Barth syndrome only, a rare mitochondrial disease. There is no completed human RCT of elamipretide for stroke, traumatic brain injury, dementia or any neurodegenerative cognitive endpoint, and across four large non-neuro RCTs (mitochondrial myopathy, heart failure, acute MI, dry AMD) it missed every primary endpoint. Its neuroprotection case is entirely preclinical: in a mouse TBI model SS-31 reversed mitochondrial dysfunction, reduced reactive-oxygen-species and attenuated neuronal apoptosis. That is a compelling mechanism, but rodent data are not human proof, and an approval for one rare disease does not make a molecule a validated neuroprotectant.
Doesn't ARA-290 have real human trial data?
Yes — but for peripheral small-fiber neuropathy, not brain neuroprotection. In randomized, double-blind, placebo-controlled work in sarcoidosis- and diabetes-associated small-fiber neuropathy, ARA-290 (cibinetide) improved neuropathy symptom and quality-of-life scores and increased objective markers of small-nerve-fiber regrowth, such as corneal nerve-fiber area and regenerating skin fibers. That is genuine Grade-B human evidence, and it is meaningfully clean because ARA-290 binds a non-erythropoietic face of erythropoietin and does not raise hemoglobin. But there is no human trial of ARA-290 for stroke, TBI or any central neurodegenerative disease, so 'brain neuroprotection' is not supported. Nerve repair in the skin or cornea is not the same as protecting the brain.
Are these peptides legal in the US in 2026?
None is an FDA-approved neuroprotectant. Cerebrolysin is not FDA-approved and has no US compounding pathway (a porcine-brain extract with no single defined active ingredient and a theoretical prion concern), so it cannot be legally prescribed or dispensed here, though it is used in 50-plus countries. Semax and ARA-290 are not FDA-approved; Semax was removed from 503A Category 2 and is scheduled for a Pharmacy Compounding Advisory Committee review on July 24, 2026 — a pending, non-binding step, not approval or clearance to compound. SS-31/elamipretide is approved only for Barth syndrome; all neuro uses are investigational. P021 is an unapproved research compound. For tested athletes, non-approved agents fall under the WADA S0 catch-all, prohibited at all times.