A watch list, not a try list
The interventions with the most to prove.
Every other ranking here sorts by how well established something is. This one asks a different question — if it worked, how much would it matter? — and sorts by the answer. That is a judgment, not a measurement, and the evidence grade travels with every row so you can see both at once.
Read this as a list of things to follow, not things to take. Ranking high here means an idea would be important if it panned out — which is the opposite of a reason to try it. Of the 10 entries assessed for it, 7 have no outcome a person could measure at all, so taking them tells you nothing whatever happens. Several are clinical-trial assets that are not purchasable, and several more have failure modes that stopping does not undo.
Potential is editorial judgment, not evidence. A high number here says an idea would matter if it worked — not that it works, not that it is safe, and not that anyone should take it. It never enters the composite grade and cannot raise one. Most entries on this page are unproven in humans, and several are sold without regulatory oversight. How this is scored →
- 9 /10
Field-defining if it translates
Rests on The most reproducible lifespan extension in mammalian biology: the NIA Interventions Testing Program extended median lifespan in genetically heterogeneous mice across three independent sites, in both sexes, and even when started in late life. The mTOR pathway it targets is conserved and central to nutrient sensing.
May not translate No human trial has ever measured a lifespan or mortality endpoint, and PEARL — the first randomised healthspan trial — missed its primary endpoints. Chronic immunosuppression, impaired glucose tolerance, and impaired wound healing are documented in transplant use, and the intermittent low-dose schedules the longevity community uses have never been shown to preserve the benefit while avoiding those effects.
Would settle it A randomised trial with a hard clinical endpoint rather than a biomarker panel.
- 9 /10ER-100 (partial epigenetic reprogramming) First-in-human Grade D · 2.9/10 Nothing you could measure Not reversible
Field-defining if it translates
Rests on Partial reprogramming is the only intervention class that has restored youthful gene-expression and methylation patterns in aged mammalian tissue, and in some models restored function (optic-nerve regeneration, improved survival in progeroid mice). If age is partly an epigenetic state, this is the class that acts on the cause rather than a downstream symptom.
May not translate The whole class carries a teratoma and dedifferentiation risk that is well documented in animals and unsolved in humans; delivery, dose, and duration control are all open problems. First-in-human dosing began in 2026 and no efficacy or safety data have been published — the ceiling here rests entirely on animal work.
Would settle it Published Phase 1 safety data, including any dedifferentiation or tumour signal.
- 7 /10
Large effect, narrow evidence
Rests on Clearing senescent cells extended healthspan and median lifespan in mice and reversed specific age-related pathologies. Early human trials in idiopathic pulmonary fibrosis and diabetic kidney disease showed condition-specific functional improvement and confirmed senescent-cell clearance in human tissue, which is a real translational step most of this field has not reached.
May not translate Human results are small, condition-specific, and unblinded in places; the intermittent dosing schedule is inferred from mouse work rather than established in people. Dasatinib is a chemotherapy agent with real toxicity, and senescent cells have beneficial roles in wound healing and tumour suppression that broad clearance may compromise.
Would settle it A randomised placebo-controlled trial in an ageing (rather than disease) population.
- 7 /10
Large effect, narrow evidence
Rests on Klotho-deficient mice show accelerated ageing across multiple organ systems, and overexpression extends lifespan — one of the cleaner single-gene demonstrations in the field. Human observational data associate higher circulating klotho with better cognitive and kidney outcomes, and a 2023 primate study reported cognitive improvement from klotho administration.
May not translate It is a large protein with no oral route and poor blood-brain-barrier penetration, so the delivery problem is unsolved. Observational human associations are heavily confounded by kidney function. No human trial of klotho administration has reported results, and nothing is purchasable — products marketed as klotho boosters have no demonstrated effect on the protein.
Would settle it A first-in-human trial of klotho administration with a pharmacokinetic readout.
- 6 /10
Real mechanism, modest ceiling
Rests on A genuinely novel mechanism — it binds cardiolipin to stabilise mitochondrial cristae rather than acting as a general antioxidant — and it earned FDA accelerated approval in 2025 for Barth syndrome, making it the first approved mitochondria-targeted therapeutic. Proof that the target class is druggable in humans.
May not translate The approval is in an ultra-rare genetic disease with roughly 150 US patients, where mitochondrial dysfunction is the primary defect. Trials in primary mitochondrial myopathy and heart failure with preserved ejection fraction were mixed to negative, which is the more relevant test for age-related mitochondrial decline. Injectable, expensive, and unavailable for any healthy-ageing use.
Would settle it A positive trial in an age-related rather than genetic mitochondrial indication.
- 6 /10
Real mechanism, modest ceiling
Rests on A positive allosteric modulator of Trk receptors — a genuinely novel mechanism aimed at amplifying endogenous neurotrophin signalling rather than replacing it — that has completed Phase 1 in humans with reported target engagement and an acceptable safety profile.
May not translate Phase 1 establishes tolerability, not benefit. The neurotrophin field has a long history of promising mechanisms failing at efficacy, and cognitive endpoints in particular have a poor track record. No efficacy data have been reported in any patient population.
Would settle it First efficacy readout in a cognitive-impairment population.
- 6 /10
Real mechanism, modest ceiling
Rests on Targets lysosomal function directly — TRPML1 agonism enhances lysosomal biogenesis and autophagic flux, and loss of proteostasis is one of the better-supported hallmarks of ageing. Acting on the disposal machinery rather than a single aggregate is a mechanistically attractive position, and it has reached first-in-human.
May not translate Everything supporting the ceiling is preclinical. Enhancing autophagy systemically has plausible downsides, including effects on tumour-cell survival, and no human efficacy or safety data have been published. First-in-human status is a starting line, not evidence.
Would settle it Published Phase 1 safety and target-engagement data.
- 6 /10
Real mechanism, modest ceiling
Rests on The first therapies to change the underlying pathology of Alzheimer's disease rather than symptoms — amyloid clearance is unambiguous on imaging, and lecanemab and donanemab both showed statistically significant slowing of decline in large Phase 3 trials, which is a genuine scientific milestone after decades of failure.
May not translate The clinical effect sizes are small and their real-world meaningfulness is contested — the differences sit near or below commonly cited thresholds for a noticeable change. ARIA (amyloid-related imaging abnormalities) causes brain oedema and haemorrhage, with deaths reported, and APOE4 homozygotes are at markedly higher risk. Cost, infusion burden, and MRI monitoring are substantial. This is proof the pathway is modifiable more than proof it is the right lever.
Would settle it Longer-term data on whether the divergence widens or plateaus.
- 5 /10
Real mechanism, modest ceiling
Rests on One of the few compounds with a specific, measurable mechanism — mitophagy induction — confirmed in human muscle biopsies, not just inferred. Human trials show improved muscle endurance and mitochondrial gene-expression signatures in middle-aged and older adults.
May not translate Effect sizes are small and concentrated in biomarkers and endurance measures rather than outcomes anyone would notice unaided. Only a minority of people convert dietary ellagitannins to urolithin A, which is the reason the supplement exists — but it also means the population-level case rests on that conversion gap mattering. Most trials are funded by the company selling it.
Would settle it An independent trial with a clinically meaningful functional endpoint.
- 5 /10
Real mechanism, modest ceiling
Rests on Extended median lifespan and, more notably, compressed morbidity in mice — the animals spent a smaller fraction of life in poor health, which is closer to what healthspan advocates actually want than lifespan alone. AKG sits at a real metabolic junction as a TCA intermediate and epigenetic cofactor.
May not translate The mouse lifespan effect was modest and, in the key study, driven substantially by female animals. Human data are limited to biomarker and epigenetic-clock changes from a small industry-associated study. Endogenous AKG is abundant, so it is unclear whether oral supplementation meaningfully changes tissue concentrations.
Would settle it An independent human trial with a functional endpoint.
- 5 /10
Real mechanism, modest ceiling
Rests on Induces autophagy through a well-characterised mechanism, extends lifespan across yeast, flies, worms, and mice, and — unusually for this field — has supporting human observational data linking higher dietary intake to lower all-cause mortality.
May not translate The observational human data carry the usual confounding by overall diet quality: spermidine-rich foods are wheat germ, legumes, aged cheese, and natto, which travel with other healthy patterns. Supplement doses are far below dietary intake in the cohorts studied, and no randomised trial has shown a clinical outcome. Gut bacteria produce substantial endogenous spermidine, so the marginal effect of supplementation is unclear.
Would settle it A randomised trial with a functional endpoint at a dose comparable to dietary intake.
- 5 /10
Real mechanism, modest ceiling
Rests on The senolytic with the best safety margin — a flavonoid found in food, with mouse work showing senescent-cell clearance and extended median and maximum lifespan even when started late in life. Currently in Mayo Clinic clinical trials, which is further than most senolytic candidates.
May not translate Oral bioavailability is poor and the doses used in mouse work are far above anything achieved by ordinary supplementation. Human trial results have not read out. The "hit and run" monthly protocol circulating in the community is extrapolated from animal dosing, not established in people.
Would settle it Readout of the Mayo Clinic senolytic trials.
- 5 /10
Real mechanism, modest ceiling
Rests on The most clinically validated peptide in this group — approved in over 30 countries as Zadaxin for hepatitis B and C and as an immune adjuvant, with randomised data in sepsis and a large body of use in immunocompromised patients.
May not translate The approvals and trials are in specific immune-compromised or infected populations, where restoring T-cell function has a clear rationale. That does not extend to immune enhancement in healthy adults, where no controlled data exist and the direction of benefit is not obvious — modulating an already-functioning immune system is as likely to be neutral or harmful.
Would settle it A controlled trial in a healthy or merely ageing (not immunocompromised) population.
- 5 /10
Real mechanism, modest ceiling
Rests on A sigma-1 receptor agonist targeting autophagy and cellular-stress handling rather than amyloid — a mechanistically distinct bet in a field where the dominant hypothesis has produced modest results. It has completed Phase 2b/3 in Alzheimer's disease.
May not translate The Phase 2b/3 results were contested: the primary analysis was widely criticised, EMA review was unfavourable, and the effect sizes claimed rest on analytical choices rather than an unambiguous separation. Sigma-1 remains an under-validated target, and the Alzheimer's field has the highest late-stage failure rate in medicine.
Would settle it An unambiguous regulatory decision or an independent confirmatory trial.
- 5 /10
Real mechanism, modest ceiling
Rests on A triple monoamine reuptake inhibitor that produced roughly double the weight loss of the anti-obesity drugs available at the time of its Phase 2, which was a striking result and is why it remains discussed.
May not translate Development stalled over cardiovascular and psychiatric tolerability — heart rate and blood pressure increases, plus mood effects consistent with its mechanism. The triple-monoamine class has a poor regulatory history for exactly these reasons. It has since been overtaken by incretins that achieve comparable or greater loss with a better safety profile, which lowers the ceiling on relevance as much as on efficacy.
Would settle it A regulatory approval in any major market.
- 5 /10
Real mechanism, modest ceiling
Rests on Heterochronic parabiosis produced real rejuvenation effects in mice across multiple tissues, and the follow-up work suggests dilution of age-elevated factors matters more than any youthful factor — which is a mechanism plasma exchange can actually act on. One small human RCT reported biological-age reductions.
May not translate Parabiosis shares an entire circulatory system continuously; a few hours of apheresis is a weak analogue. The human evidence is one small trial using epigenetic-clock endpoints that are themselves unvalidated as outcome surrogates. The procedure is invasive, expensive, requires clinical supervision, and carries citrate-reaction and line risks.
Would settle it A randomised trial with a functional endpoint rather than a biological-age clock.
- 5 /10Dihexa (N-hexanoic-Tyr-Ile-(6) aminohexanoic amide) Preclinical Grade D · 3.1/10 Nothing you could measure Not reversible
Real mechanism, modest ceiling
Rests on An angiotensin IV analogue reported to be orders of magnitude more potent than BDNF at promoting synaptogenesis in preclinical models, with striking results in animal models of cognitive impairment. If a fraction of that potency translated it would be a significant compound.
May not translate There is no human data of any kind. A potent synaptogenic agent is exactly the class where uncontrolled growth is a serious theoretical concern, and the HGF/c-Met pathway it acts through is implicated in oncogenesis. It is sold as a research chemical with no safety file whatsoever, which is a materially different situation from an unproven drug in development.
Would settle it Any formal preclinical toxicology package, let alone a human trial.
- 4 /10
Mechanistic story only
Rests on An ACTH(4-10) analogue with a coherent mechanism — it raises BDNF and NGF expression in animal models — and decades of Russian clinical use in stroke and cognitive indications, including registration as a medicine there.
May not translate The clinical literature is almost entirely Russian-language, with trial designs and reporting standards that have not been independently verified, and Western replication is essentially absent. Intranasal peptide delivery to the brain is itself contested. This is a case where the volume of use is high and the quality of evidence supporting it is low.
Would settle it An independent Western trial with a validated cognitive endpoint.
- 4 /10
Mechanistic story only
Rests on The best-characterised GHRH analogue, previously FDA-approved for paediatric GH deficiency — so unlike most peptides in this space it has a real regulatory and safety file behind it.
May not translate The approval was for diagnosed deficiency in children, which says little about healthy adults seeking enhancement. It was withdrawn from the US market for commercial rather than safety reasons, and the healthy-adult use has no controlled outcome data. The GH-longevity relationship points the wrong way.
Would settle it A controlled adult trial with a functional rather than hormonal endpoint.
- 4 /10
Mechanistic story only
Rests on A CD38 inhibitor, which gives it a specific and interesting mechanism — CD38 consumes NAD+ and rises with age, so inhibiting it is a mechanistically distinct route to the same target the NAD+ precursors chase, potentially a more efficient one.
May not translate The CD38 work is preclinical and the concentrations used in vitro are far above what oral apigenin achieves; bioavailability is poor. Human data are essentially limited to sleep and anxiolytic use at chamomile-derived doses, which is a different claim entirely. No human study has shown a change in NAD+ from apigenin.
Would settle it A human study measuring NAD+ after apigenin dosing.
- 4 /10
Mechanistic story only
Rests on A mitochondrial-derived peptide — a genuinely novel signalling class — that improved exercise capacity in old mice to the level of young animals, and whose circulating levels rise with exercise in humans, suggesting it is part of a real physiological pathway rather than a pharmacological curiosity.
May not translate Everything above is mouse work plus a human correlation. The first registered human trial began recruiting in 2026 and has no results. Injectable, unapproved, with no established human dose, and the correlation between exercise and circulating MOTS-c does not establish that administering it reproduces the effect.
Would settle it First published human trial results.
- 4 /10
Mechanistic story only
Rests on A selective ghrelin-receptor agonist that raises GH with less cortisol and prolactin disturbance than earlier secretagogues — a genuinely cleaner pharmacological profile within its class.
May not translate Same ceiling problem as the whole GH-secretagogue class: the demonstrated effect is on a hormone level, not an outcome. Development was discontinued after trials in post-operative ileus, so the compound has been tested in humans and not pursued. Nothing establishes benefit in healthy adults.
Would settle it Any controlled trial with a functional endpoint in a non-deficient population.
- 4 /10
Mechanistic story only
Rests on The SARM with the best human data: placebo-controlled Phase 1 work in healthy men showed dose-dependent lean-mass gain over three weeks with a favourable safety profile in that window. The tissue-selectivity premise — anabolic effect with less prostate and androgenic activity — is real in animal models.
May not translate The human evidence is short-duration Phase 1 in small numbers, and clinical development did not proceed to efficacy trials in any indication. Suppression of endogenous testosterone is consistent and dose-dependent, liver enzyme elevations are documented, and long-term cardiovascular and endocrine effects are entirely unstudied. Product sold online frequently does not contain what the label claims.
Would settle it A trial of longer than 12 weeks with endocrine recovery data.
- 4 /10
Mechanistic story only
Rests on A GHRH analogue that reliably raises endogenous GH and IGF-1 pulses in humans — the pharmacology is real and measurable, not hypothetical, and the DAC version extends half-life enough for infrequent dosing.
May not translate Raising GH is not the same as benefiting from it. Human data stop at the hormone level; no trial has shown a body-composition, functional, or healthspan outcome in healthy adults. GH excess causes insulin resistance, oedema, carpal tunnel, and is epidemiologically associated with worse rather than better longevity — the long-lived human and animal phenotypes tend to have *lower* GH/IGF-1 signalling.
Would settle it A controlled trial with a functional endpoint in healthy adults.
- 4 /10
Mechanistic story only
Rests on Emerged from heterochronic parabiosis as a candidate youthful circulating factor, with initial reports of cardiac and skeletal-muscle rejuvenation in aged mice — a striking result that drove much of the young-blood research programme.
May not translate The central findings are contested: independent groups reported the opposite direction of effect and identified assay problems that confounded the original measurements, including cross-reactivity with the closely related myostatin. This is the clearest case in the database of a high-profile preclinical result that did not replicate, and the ceiling is discounted accordingly.
Would settle it Resolution of the assay and replication dispute in independent hands.
- 4 /10
Mechanistic story only
Rests on Reported dopaminergic and neurotrophic effects in animal models, including dopamine-neuron protection and increases in dendritic complexity — an unusual combination of stimulant-like and neurotrophic activity in one molecule.
May not translate Rodent work only, from a small number of groups. Beta-carbolines are a chemically promiscuous class with MAO-inhibitory activity and interaction potential, and no human pharmacokinetic or safety data exist. Sold as a research chemical with no oversight of identity or purity.
Would settle it Independent replication plus a first human pharmacokinetic study.
- 4 /10
Mechanistic story only
Rests on The C-terminal tripeptide of alpha-MSH, with consistent anti-inflammatory effects in colitis, wound-healing, and dermatitis models, and an appealing property: it retains anti-inflammatory activity without the pigmentation and cardiovascular effects of the full alpha-MSH molecule.
May not translate Essentially all of that is rodent and cell work. Human evidence is close to absent, oral stability and absorption are unresolved, and the IBD and skin uses it is sold for have no controlled human data at all.
Would settle it A first controlled human trial in an inflammatory indication.
- 4 /10
Mechanistic story only
Rests on An unusually broad and consistent body of rodent work on tendon, ligament, muscle, and gut healing, spanning decades from the originating group — enough repetition that the effect in rodents is not seriously in doubt.
May not translate Almost the entire literature originates from a single research group, which is the classic setup for a result that does not survive independent replication. There are no published human randomised trials at all. The FDA has flagged immunogenicity and peptide-impurity concerns for compounded product, and the peptide sold online is of unverified identity and purity.
Would settle it A published human randomised trial from an independent group.
- 4 /10
Mechanistic story only
Rests on Thymosin beta-4 is a real endogenous actin-sequestering protein with credible roles in cell migration and tissue repair, and RegeneRx carried it into Phase 2 trials for dry eye, stroke, and diabetic ulcers — further than most peptides in this category reach.
May not translate Those Phase 2 programmes were unconvincing and largely stalled, which is evidence against rather than pending. TB-500 is a fragment, and vendors frequently supply full TB-4 instead, so what people take is often not what was studied. No human data exist for the musculoskeletal-repair use the peptide is actually bought for.
Would settle it A controlled trial in tendon or muscle injury, the indication it is used for.
- 3 /10
Mechanistic story only
Rests on The original nootropic, with a very large accumulated literature, decades of European clinical use, and a benign safety record — the most-studied compound in its class by a wide margin.
May not translate The volume of literature is not matched by its quality: results in healthy adults are weak and inconsistent, the better-designed trials are in cognitive impairment and dyslexia rather than enhancement, and Cochrane-style syntheses have not supported a cognitive-enhancement claim. Decades of study without a clear positive is itself informative — this is closer to a settled negative than an open question.
Would settle it None pending — the enhancement question has largely been asked and answered.
- 3 /10
Mechanistic story only
Rests on A dipeptide with a proposed BDNF and NGF-mediated mechanism, animal data on memory consolidation, and Russian clinical use — active at far lower doses than the racetams it is grouped with.
May not translate Human evidence is limited to small Russian studies, mostly in cognitive impairment rather than healthy users. No Western trial exists. Reported effects in healthy adults are largely self-reported and would be indistinguishable from expectancy in an unblinded setting.
Would settle it A placebo-controlled trial in healthy adults.
- 3 /10
Mechanistic story only
Rests on A tuftsin analogue registered in Russia as an anxiolytic, with animal work suggesting GABAergic and BDNF-related effects and, unusually for an anxiolytic, no sedation or dependence signal.
May not translate The same problem as Semax and more so: the evidence base is one national research tradition, small, and not independently replicated. There are no Western controlled trials, and the anxiolytic field has an especially poor record of small unblinded results surviving proper blinding.
Would settle it An independent placebo-controlled trial.
- 3 /10
Mechanistic story only
Rests on A plausible mechanism — transient opioid-receptor blockade producing compensatory endorphin upregulation, plus TLR4-mediated glial anti-inflammatory effects — and a very benign safety profile at low dose, with early small crossover trials suggesting benefit in fibromyalgia and Crohn's.
May not translate The first long-term properly powered test, a 12-month randomised placebo-controlled trial in fibromyalgia, found no clinically meaningful benefit on pain or any secondary outcome. That is the strongest evidence available for the indication LDN is most used for, and it is negative. The remaining signals are small, unblinded, or in conditions where placebo response is high.
Would settle it A properly powered trial in an indication other than fibromyalgia.
- 3 /10
Mechanistic story only
Rests on Strong anabolic and neuroprotective signals in animal models and a high anabolic-to-androgenic ratio in preclinical work, which is the property the entire SARM class is premised on.
May not translate Clinical development was discontinued, and the FDA has issued warnings citing drug-induced liver injury reports associated with RAD-140 products. There is no published human efficacy trial. A compound whose sponsor stopped developing it and whose regulator has issued a safety communication has a materially lower ceiling than one merely awaiting data.
Would settle it None pending — development discontinued.
- 3 /10
Mechanistic story only
Rests on A direct AMPK activator that increased endurance in sedentary mice — the original "exercise in a pill" result — and AMPK is a well-validated node in the nutrient-sensing pathways that matter for ageing.
May not translate Poor oral bioavailability means it must be injected in large quantities, which is impractical and expensive. Human trials have been limited and unimpressive, AMPK activation has effects on tumour metabolism that cut both ways, and the mouse endurance result has not translated. WADA-prohibited.
Would settle it A human trial with a functional endurance endpoint.
- 3 /10
Mechanistic story only
Rests on A genuine component of human innate immunity with broad antimicrobial and immunomodulatory activity in vitro, and a plausible role in wound healing.
May not translate LL-37 is strongly context-dependent in the body and has been implicated as a driver of pathology in psoriasis, rosacea, and lupus, where it promotes autoimmune interferon responses. Administering more of it is not obviously desirable. No human trials support supplemental use, and injecting an endogenous immune peptide is exactly the situation where preclinical enthusiasm most often reverses.
Would settle it A first-in-human safety study clarifying the autoimmunity risk.
- 3 /10
Mechanistic story only
Rests on REV-ERB agonists reported to produce "exercise-mimetic" effects in mice — increased mitochondrial content and running endurance without training — which is why the compound acquired its reputation.
May not translate A 2020 study found the headline endurance effects persisted in REV-ERB knockout mice, meaning the observed effects were substantially off-target rather than mechanism-driven. Oral bioavailability is very poor, so human exposure from oral dosing is minimal, and there is no human data at all. This is the clearest case in the database of a preclinical result whose stated mechanism did not hold up.
Would settle it Resolution of the off-target findings; no human programme exists.
- 2 /10
Little to support a ceiling
Rests on A long Russian clinical tradition reports mortality and functional benefits, and there is a proposed telomerase-activation mechanism.
May not translate Essentially the entire evidence base comes from one institute, largely outside Western peer review, with trial designs and data that have not been independently verified. Telomerase activation is not obviously desirable — it is a recognised cancer-risk pathway, and no long-term oncological follow-up exists. The mechanism is asserted more than demonstrated.
Would settle it Independent replication outside the originating research tradition.
- 1 /10
Little to support a ceiling
Rests on Rodent studies report increased testosterone, and it acquired a following through podcast recommendation rather than through any clinical result.
May not translate No published human efficacy trial of any kind. Rodent studies additionally reported testicular toxicity at higher doses, so the animal literature that generated the interest also contains a safety signal. There is no characterised active compound, no standardised extract, and no human dose rationale.
Would settle it Any published human trial.
Coverage is deliberately partial. An entry appears here only where a real basis and a real counterweight can both be written; most of the database carries neither and is better served by its evidence grade alone. The full ranked database is at /rankings/. This page sorts by how much something would matter and carries no timetable; for what is due to report soon and what is just entering human testing, see /upcoming/.
Get the weekly update
A concise record of what changed, why it changed, and what is still uncertain.
Signup is handled by Beehiiv. See our privacy notice.