Oxygen has two roles in the body. The first is substrate: it accepts electrons at the end of the mitochondrial electron transport chain, enabling oxidative phosphorylation and ATP production. The second is signal: oxygen levels regulate gene expression, cell behavior, and vascular tone through pathways involving hypoxia-inducible factors, reactive oxygen species, and reactive nitrogen species.

Standard clinical medicine focuses almost entirely on the substrate role. Hyperbaric oxygen therapy, HBOT, is interesting precisely because it exploits both. By transiently driving tissues into hyperoxia, it does something beyond filling the ATP production queue: it activates signaling cascades whose downstream effects persist long after the session ends.

This distinction is the spine of this article. The substrate story is settled physics, fully understood, clinically proven for specific indications, and uncontroversial. The signaling story is real biology, mechanistically established in cell culture and animal models, with genuine clinical translation to some outcomes and contested translation to others. The longevity and cognitive claims that have generated most of the popular interest sit in a third tier: biologically plausible, built on mechanisms that are real, but resting on a clinical trial base that is small, often unblinded, and concentrated in a single research ecosystem.

The reader should know which tier they are in at every point. This article marks each one explicitly.

Tier 1: Settled physiology

Henry’s law states that the amount of a gas dissolved in a liquid is proportional to the partial pressure of that gas above the liquid. At sea level, breathing room air, the partial pressure of oxygen is approximately 160 mmHg, and the amount dissolved directly in plasma is about 0.3 vol%: a trivial fraction of the total oxygen carried, since hemoglobin handles the rest. This is not a problem under normal circumstances because hemoglobin is nearly saturated already.

Under hyperbaric conditions, the physics change substantially. At 2.4 ATA breathing 100% oxygen, plasma-dissolved oxygen can exceed 6 vol%, roughly twenty times the sea-level baseline. This is a clinically meaningful quantity. It is enough to sustain tissue oxygenation without red blood cells at all: the plasma alone can deliver sufficient oxygen if circulation is intact.^[1]

This mechanism, dissolved oxygen bypassing hemoglobin and reaching ischemic tissue through plasma diffusion, is the physical basis for the strongest HBOT indications. In decompression sickness, nitrogen bubbles obstruct microvascular flow and must be physically compressed; the additional oxygen loading accelerates bubble reabsorption and maintains tissue viability during the process. In carbon monoxide poisoning, hemoglobin is occupied by CO, effectively removing its oxygen-carrying capacity; driving plasma O2 to hyperbaric levels maintains tissue oxygenation through a route that does not depend on hemoglobin at all. In diabetic foot ulcers and other ischemic wounds, the wound margin is hypoxic not because of CO or nitrogen but because local vascular supply is insufficient; increasing dissolved plasma oxygen reaches tissue through diffusion across shorter distances than it would at normal pressure.

None of this is hypothesis. The Henry’s law mechanism is physics. The clinical benefit in these specific indications follows from it in a straightforward chain with support from randomized controlled trials. The physiology is understood. The delivery engineering is understood. This is the settled tier.

Tier 2: Oxygen as signal

Here the picture becomes more interesting and more complicated.

The cell biology of oxygen sensing is well established. The 2019 Nobel Prize in Physiology or Medicine went to William Kaelin Jr., Peter Ratcliffe, and Gregg Semenza for characterizing the HIF pathway: specifically, how cells sense and adapt to oxygen levels. HIF-1alpha is a transcription factor that is continuously produced in cells and, under normal oxygen conditions, continuously degraded by a family of prolyl hydroxylase enzymes that require oxygen to function. When oxygen falls, prolyl hydroxylase activity drops, HIF-1alpha accumulates, and a large gene expression program activates, including genes for VEGF (vascular endothelial growth factor), erythropoietin, and metabolic adaptation enzymes.

Hyperoxia drives this in the opposite direction: elevated oxygen accelerates HIF-1alpha degradation. But what happens at reoxygenation, the transition from hyperoxia back to normal oxygen tension, is where HBOT’s signaling effects begin to make biological sense.

During and after a hyperbaric session, the reoxygenation transition generates a burst of reactive oxygen species and reactive nitrogen species. ROS and RNS are not simply damage signals; at controlled concentrations they are genuine second messengers. The specific molecular events documented in the literature include VEGF upregulation, activation of endothelial nitric oxide synthase, and mobilization of CD34-positive stem and progenitor cells from bone marrow into circulation. This last effect has been characterized in human subjects: Thom and colleagues measured CD34+ cell counts before and after HBOT sessions and found approximately eight-fold increases in circulating progenitor cells compared to controls breathing room air at normal pressure.^[2]

The proposed integrating framework is the “hyperoxic-hypoxic paradox,” articulated by Hadanny and Efrati: intermittent hyperoxia followed by return to normoxia creates a redox swing that mimics, at the signaling level, the effects of hypoxic preconditioning.^[3] Hypoxic preconditioning is a well-characterized phenomenon: brief periods of ischemia before a larger ischemic event reduce subsequent damage, partly through HIF pathway activation and partly through mitochondrial conditioning. The HBOT hypothesis is that the signaling consequences of the hyperoxia-to-normoxia transition are functionally analogous, achieved from the opposite direction.

The mechanistic claim here deserves a clear statement of confidence. The oxygen sensing pathway is settled Nobel-prize-level biology. The ROS and RNS second-messenger role is well established in vitro and in animal models. Stem cell mobilization by HBOT has been demonstrated in humans. These are not speculative claims.

What is uncertain is the magnitude of these effects in humans under clinical protocols, how much the downstream signal translates to measurable clinical outcomes, and which outcomes are meaningfully affected and by how much. The mechanism is real. The clinical consequence of the mechanism, in specific human populations at specific protocols, is where the science is still resolving.

A clarification on a popular metaphor: the claim sometimes made is that “oxygen escaping as you exit the chamber is what provides the signal.” This is a loose description of the reoxygenation redox swing. The actual biology is more specific: the transition from elevated partial pressure to normal partial pressure changes redox status in cells, generating ROS and RNS that trigger transcriptional responses including HIF pathway re-activation. The metaphor is not wrong in pointing at the transition as the key event, but it obscures the actual mechanism, which is electrochemical, not simply about gas escaping.

Tier 3: Clinical outcomes with RCT-grade evidence

The Undersea and Hyperbaric Medical Society maintains an approved indications list, updated periodically, representing the consensus of hyperbaric medicine on where evidence is sufficient to justify treatment. As of the most recent revision, approved indications include: decompression sickness, carbon monoxide poisoning, arterial gas embolism, clostridial myonecrosis (gas gangrene), crush injury and acute traumatic ischemia, necrotizing soft tissue infections, refractory osteomyelitis, osteoradionecrosis, soft tissue radionecrosis, late radiation tissue injury, diabetic foot ulcers, compromised grafts and flaps, and acute peripheral arterial insufficiency.

The evidence quality across this list is not uniform, and the Cochrane reviews are instructive in their precision.

Carbon monoxide poisoning represents the strongest single trial. Weaver and colleagues published a double-blind, randomized controlled trial in the New England Journal of Medicine in 2002 showing that three HBOT sessions within 24 hours of CO exposure reduced cognitive sequelae at six weeks by approximately half compared to normobaric oxygen.^[4] The blinding was credible: both groups wore similar masks, breathed through similar apparatus. This is the best trial in the HBOT literature.

For diabetic foot ulcers, the Cochrane systematic review (2015) concluded there was “some benefit” for wound healing and reduction in major amputation rates, specifically based on three randomized trials totaling approximately 120 patients, but rated the evidence as low-certainty due to small sample sizes, methodological limitations, and clinical heterogeneity across wound types and patient populations.^[5] The benefit signal is present; the quantitative confidence is limited.

For late radiation tissue injury and osteoradionecrosis, the evidence base includes several randomized trials with consistent benefit signals for wound healing endpoints, though most are small. The mechanistic rationale is strong: radiation-damaged tissue has depleted vascularity and limited oxygen delivery; restoring dissolved oxygen bypasses this limitation and appears to facilitate collagen synthesis and angiogenesis.

The practical interpretation of this tier: for the established indications, the clinical evidence justifies the therapy at a level that meets standard evidence-based medicine criteria. For some indications that justification is robust; for others it rests on low-certainty evidence that is nonetheless better than anything available for the longevity and cognitive claims.

Tier 4: The longevity and cognitive claims

This is the exciting tier and the most important one to read carefully.

The biological logic connecting HBOT to longevity-relevant endpoints is coherent. If HBOT mobilizes stem and progenitor cells, upregulates VEGF and promotes angiogenesis, and reduces chronic inflammation through ROS-mediated transcriptional effects, then downstream consequences could plausibly include tissue repair, senescent cell clearance, and maintenance of cognitive function in aging tissue. This chain of reasoning is not invented; it follows from the signaling biology in Tier 2.

The problem is that “plausibly could follow” is not the same as “has been demonstrated to follow in controlled human trials.” Here is what the trial evidence actually says.

Telomere lengthening and senescent cell reduction. Hachmo and colleagues published results in Aging (Albany NY) in 2020 reporting that 60 HBOT sessions over 90 days produced a 20-38% increase in telomere length and 11-37% reduction in circulating senescent T-cells in a cohort of 35 healthy aging adults.^[6] The effect sizes are striking, larger than those reported for most pharmaceutical or lifestyle interventions studied in this context. The limitations are equally striking: no sham-control group, no blinding, single cohort. The study cannot distinguish HBOT effects from regression to the mean, expectation effects, seasonal variation, or any other time-correlated factor. The research group is from the Sagol Center for Hyperbaric Medicine and Research, affiliated with Aviv Scientific, a commercial HBOT provider. Independent replication of these specific findings does not exist as of mid-2026.

Cognitive gains in healthy aging. The same research group has published a series of studies showing improvements in processing speed, executive function, and attention following HBOT in adults over 65. The individual study sample sizes range from approximately 35 to 73 participants. Most lack sham controls. The cognitive improvements, where measured against baseline, are statistically significant, but the absence of an adequately controlled comparison group means the magnitude of HBOT-specific effect cannot be separated from test-retest learning effects, which are substantial in cognitive test batteries. Cognitive test scores tend to improve on second administration simply from familiarity with the test format.

Post-stroke and post-TBI neuroplasticity. Efrati and colleagues published results in PLoS ONE in 2013 showing improvements in neurological function in chronic stroke patients, some of them years post-event, following HBOT.^[7] The proposed mechanism, reactivation of “dormant” neurons in the penumbral zone around infarct tissue, is biologically plausible: tissue adjacent to an infarct can remain metabolically suppressed but structurally intact for years, and increasing oxygen delivery could, in principle, restore some function. The trials are small and the evidence is preliminary; they do not constitute a clinical standard of care. Larger independent trials are ongoing.

Fibromyalgia. Efrati and colleagues published results in PLoS ONE in 2015 from a randomized crossover trial of 60 fibromyalgia patients, including a sham-control arm using normobaric air.^[8] Tender point counts, pain thresholds, and quality-of-life measures improved significantly in the HBOT arm. This study has better methodology than the telomere and cognitive trials: a crossover design with a control condition, even if the sham arm’s physiological inertness is debatable at 1.3 ATA air. It is the most credible longevity-adjacent HBOT trial. Replication by independent groups has not yet appeared.

The honest summary of this tier: the effects are biologically plausible, some individual studies show striking results, and the mechanisms in Tier 2 provide a coherent rationale for why they could exist. None of these outcomes yet rests on a trial base sufficient to call them established. They are hypotheses with preliminary positive signals.

Caveats: foregrounded, not footnoted

These limitations are not peripheral qualifications to an otherwise solid case. They are central to an accurate assessment of where the field stands.

The sham-control problem. HBOT trials face an inherent blinding challenge that most drug trials do not. Entering a pressurized chamber is a distinctive sensory experience: the ears pop, the air feels different, the environment is unmistakably unusual. Participants in uncontrolled studies know they are receiving the treatment, and expectation effects on subjective outcomes (pain, quality of life, cognitive self-report) are substantial. Some trials use a sham condition of 1.3 ATA breathing enriched air or normal air. But 1.3 ATA is not physiologically inert: it raises plasma-dissolved oxygen above baseline, and dissolved oxygen at 1.3 ATA provides some signaling effects, meaning the “placebo” is not a true placebo. Without a credible sham, regression to the mean and expectation cannot be ruled out for any outcome that is self-reported or that improves naturally over time.

Conflict of interest. The proportion of positive longevity and cognitive HBOT findings originating from a single ecosystem is notable and must be stated plainly. Shai Efrati, the Sagol Brain Institute at Shamir Medical Center, and Aviv Scientific, the commercial HBOT clinic network, account for a disproportionate share of the published longevity and cognitive results. This does not make those results false. Researchers with strong personal and institutional commitment to a hypothesis do sometimes produce correct positive results. But independent replication, by researchers without financial or reputational stakes in the outcome, is the standard mechanism by which such results are either validated or revised. That replication has not yet occurred at meaningful scale for the longevity endpoints.

Protocol heterogeneity. “HBOT” is not a single therapy. The parameter space includes pressure (1.3 to 2.5 ATA or higher), oxygen concentration (100% O2 vs. enriched air vs. room air), session duration (60 to 90 minutes typically), session frequency, and total number of sessions (20 to 60 or more). Results from one protocol do not transfer to another. The specific Efrati/Sagol longevity protocol uses 60 sessions at 2 ATA, 100% oxygen, 90 minutes, with five-minute air breaks every 20 minutes. Soft-chamber consumer devices at 1.3 ATA breathing ambient air are a fundamentally different intervention: they raise ambient pressure modestly, they do not deliver 100% oxygen, and the plasma O2 increase is minimal. Studies conducted at 2 ATA breathing 100% O2 do not validate the experience of 1.3 ATA ambient-air soft chambers. The popular n=1 HBOT community often runs protocols whose relationship to any studied protocol is unclear.

The hormetic window. Oxygen is hormetic: beneficial at the right dose and toxic above it. The same ROS that mediates beneficial signaling at moderate concentrations causes lipid peroxidation, DNA damage, and protein oxidation at high concentrations. Two specific toxicity syndromes are clinically established. Pulmonary oxygen toxicity (the Lorrain Smith effect) develops with prolonged exposure to elevated oxygen partial pressures: early symptoms include tracheal irritation and cough, with alveolar damage at higher exposures or longer durations. CNS oxygen toxicity manifests as grand mal seizures and occurs at partial pressures above approximately 1.6 ATA O2 in susceptible individuals. Clinical HBOT protocols are designed to stay within the therapeutic window and are supervised by trained staff specifically because of these risks. Consumer soft-chamber devices at 1.3 ATA ambient air do not approach this risk profile. Clinical HBOT at 2.0 to 2.4 ATA does, which is why it is delivered in clinical settings.

Claims that exceed the evidence. Several popular framings of HBOT benefits are not supported by the trial literature and should be flagged. The claim that HBOT produces “hormonal balance” is not supported by any rigorous RCT. The claim that HBOT produces “neurotransmitter balance” is similarly unsupported at the clinical trial level, though mechanistic pathways through which brain oxygenation could affect neurotransmitter metabolism exist. These claims are speculative. They may be worth investigating but should not be presented as established benefits.

What the evidence actually supports

A practical summary, organized by confidence level:

High confidence, RCT-grade support: Decompression sickness, carbon monoxide poisoning, and arterial gas embolism, where the Henry’s law mechanism is central and the clinical evidence is strong. Osteoradionecrosis and late radiation tissue injury, with consistent benefit signals across multiple trials.

Moderate confidence, positive signal with methodological limits: Diabetic foot ulcers, where Cochrane reviews find benefit but rate evidence as low-certainty. Problem wounds and compromised flaps, where benefit is biologically expected and clinically observed but trial quality varies.

Preliminary signal, independent replication absent: Telomere lengthening, senescent cell reduction, cognitive gains in healthy aging, post-stroke neuroplasticity, and fibromyalgia. Biologically plausible, mechanistically supported, published positive results from one ecosystem, not yet independently replicated.

Not supported: General claims about hormonal or neurotransmitter balance, claims extrapolated from clinical HBOT protocols to consumer soft-chamber devices, and any claim to established longevity benefit.

---

The loading is physics. The signaling is real biology. The longevity and cognitive claims are biologically plausible but rest on small, mostly unblinded, often conflicted trials. That does not make them wrong. It makes them bets. Worth tracking, worth n=1 testing against predefined biomarkers, not yet worth treating as established.

References

1. Leach RM, Rees PJ, Wilmshurst P. (1998). Hyperbaric oxygen therapy. BMJ, 317(7166), 1140–1143. https://pubmed.ncbi.nlm.nih.gov/9784458/
2. Thom SR, Bhopale VM, Velazquez OC, Goldstein LJ, Thom LH, Buerk DG. (2006). Stem cell mobilization by hyperbaric oxygen. American Journal of Physiology: Heart and Circulatory Physiology, 290(4), H1378–H1386. https://pubmed.ncbi.nlm.nih.gov/16299259/
3. Hadanny A, Efrati S. (2020). The hyperoxic-hypoxic paradox. Biomolecules, 10(6), 958. https://pubmed.ncbi.nlm.nih.gov/32599875/
4. Weaver LK, Hopkins RO, Chan KJ, et al. (2002). Hyperbaric oxygen for acute carbon monoxide poisoning. New England Journal of Medicine, 347(14), 1057–1067. https://pubmed.ncbi.nlm.nih.gov/12362006/
5. Kranke P, Bennett MH, Martyn-St James M, Schnabel A, Debus SE, Weibel S. (2015). Hyperbaric oxygen therapy for chronic wounds. Cochrane Database of Systematic Reviews, 2015(6), CD004123. https://pubmed.ncbi.nlm.nih.gov/26106870/
6. Hachmo Y, Hadanny A, Abu Hamed R, et al. (2020). Hyperbaric oxygen therapy increases telomere length and decreases immunosenescence in isolated blood cells: a prospective trial. Aging (Albany NY), 12(22), 22445–22456. https://pubmed.ncbi.nlm.nih.gov/33206581/
7. Efrati S, Fishlev G, Bechor Y, et al. (2013). Hyperbaric oxygen induces late neuroplasticity in post stroke patients: randomized, prospective trial. PLoS ONE, 8(1), e53716. https://pubmed.ncbi.nlm.nih.gov/23335971/
8. Efrati S, Golan H, Bechor Y, et al. (2015). Hyperbaric oxygen therapy can diminish fibromyalgia syndrome: prospective clinical trial. PLoS ONE, 10(5), e0127012. https://pubmed.ncbi.nlm.nih.gov/25950267/