The runner’s high is usually described as one feeling caused by one chemical. The experience is less orderly. Some runners report euphoria. Others notice quiet, reduced anxiety, muted pain, altered time or an effortless urge to continue. Many finish a long run without anything they would call a high.
For decades, endorphins supplied the standard explanation. Exercise can raise beta-endorphin in the bloodstream, and opioids are closely associated with pain relief and reward. But circulating beta-endorphin does not readily cross the blood–brain barrier. A blood increase cannot, by itself, explain a feeling generated inside the brain.
This is an explanation of research, not medical or exercise advice. No duration or intensity guarantees euphoria, and pushing harder simply to chase it can add risk without producing the experience.
Endocannabinoids now have the strongest claim to a primary role, especially in euphoria and reduced anxiety. Even that is not the whole story. A 2026 review in The Neuroscientist describes an interacting “neurochemical orchestra” in which the endocannabinoid system is prominent but not alone.
Endorphins arrived at exactly the right historical moment
Scientific descriptions of runner’s high appeared in the late 1970s, shortly after researchers identified endogenous opioid peptides. “Endorphin” itself combines endogenous and morphine. The name seemed almost to contain the answer.
Researchers then observed that prolonged or strenuous exercise increased beta-endorphin in blood. The sequence was intuitively satisfying: running caused the body to release a natural opioid, the opioid reduced pain, and the runner felt pleasure.
Yet the measurements left out the most important location. Beta-endorphin in peripheral blood is a relatively large, water-soluble peptide. The blood–brain barrier tightly regulates passage from circulation into the central nervous system, and beta-endorphin does not cross it readily.
The barrier does not make a peripheral rise meaningless. Circulating beta-endorphin can act elsewhere in the body, and indirect signals can affect the brain. It does mean that a blood sample cannot be treated as a reading of opioid activity inside the brain.
That distinction weakened the familiar mechanism. It did not prove that every part of the opioid system was absent from exercise.
The brain can release opioids without importing them from blood
Opioid peptides and receptors exist within the central nervous system. Locally released brain opioids do not need to cross the blood–brain barrier. This leaves room for exercise to recruit central opioid signalling even if circulating beta-endorphin is a poor messenger from blood to brain.
A 2008 positron-emission tomography study provided evidence for that possibility. Ten athletes were scanned at rest and after two hours of endurance running, covering an average of about 21.5 kilometres.
After running, opioid-receptor availability fell in several prefrontal and limbic regions. That pattern is consistent with endogenous opioids occupying receptors. Greater changes in several regions were associated with higher euphoria ratings.
The experiment was small, involved an unusually long run and measured association rather than necessity. It showed that central opioid activity can accompany a high. It could not establish that the high would disappear if opioid receptors were blocked.
This is why two statements can both be true. The circulating-endorphin explanation is inadequate, while opioids produced inside the brain may still form part of the exercise response.
The naltrexone experiment asked whether opioids were necessary
The sharper human test arrived in 2021. In a double-blind, randomised, placebo-controlled study, researchers gave regular endurance exercisers either naltrexone, which blocks opioid receptors, or a placebo.
Sixty-three participants completed a 45-minute moderate treadmill run and a 45-minute walking comparison. Researchers measured euphoria, anxiety and circulating concentrations of the endocannabinoids anandamide, known as AEA, and 2-arachidonoylglycerol, or 2-AG.
Running produced the expected pattern. Compared with walking, it increased reported euphoria, reduced anxiety and raised AEA and 2-AG. The crucial result was that naltrexone did not prevent the rise in euphoria or the reduction in anxiety.
The share of participants reporting a subjective runner’s high was also comparable in the naltrexone and placebo groups. Under those conditions, opioid-receptor activation was not required for the two emotional features the study could measure.
This does not demonstrate that opioids contribute nothing. Naltrexone did not erase every chemical response to running, and the study focused on euphoria and anxiety rather than every form of pain relief, stress regulation or reward. It is evidence against necessity, not evidence of total absence.
Endocannabinoids have a more plausible route into the brain
Anandamide and 2-AG are lipid signalling molecules made by the body. They interact with cannabinoid receptors, principally CB1 and CB2, as well as several other molecular targets. Their fat-soluble chemistry allows them to cross biological barriers more readily than circulating beta-endorphin.
The comparison with cannabis is useful but incomplete. THC can activate some of the same cannabinoid receptors, particularly CB1. The body is not manufacturing cannabis. It is making short-lived messengers on demand, controlling their location and rapidly breaking them down.
The endocannabinoid system was not discovered until the 1990s, after the endorphin explanation had already settled into popular culture. Once researchers could measure AEA and 2-AG, endurance exercise repeatedly appeared to engage the system.
A systematic review of human exercise studies examined 21 eligible articles. Fourteen of 17 studies of acute exercise detected an increase in endocannabinoids. Anandamide produced the more consistent result; findings for 2-AG varied more across samples and exercise conditions.
That pattern is stronger than a single experiment, but it remains largely correlational in people. A molecule can rise alongside mood without being the sole cause of it. Exercise intensity, duration, fitness, sample timing and laboratory methods can all change what appears in blood.
Real-world work is beginning to extend the laboratory evidence. In a 2024 study of a 60-minute outdoor run, all 16 participants reported better mood afterwards, while AEA and 2-AG both increased. The study had no non-running control group and was far too small to establish causality, but its direction matched the treadmill findings.
The strongest cannabinoid receptor test was performed in mice
Receptor blockade can move the evidence from “these things rose together” towards “this pathway was required.” The most informative cannabinoid-blocking experiment was conducted in mice.
In a 2015 study in PNAS, wheel running raised both beta-endorphin and anandamide. The researchers then used drugs and genetically altered mice to interrupt cannabinoid or opioid signalling.
After running, the mice showed less anxiety-like behaviour and lower sensitivity to pain. Blocking cannabinoid receptors removed those effects. Blocking opioid receptors did not. More detailed experiments linked the anxiety result to CB1 receptors on forebrain GABAergic neurons and the pain result to peripheral CB1 and CB2 signalling.
The species boundary is fundamental. A mouse can reveal whether it explores an exposed area or reacts to a painful stimulus. It cannot tell researchers whether it feels euphoric, experiences flow or notices time differently.
The human evidence has the opposite limitation. People can report euphoria, but researchers have not repeated the treadmill trial with a suitable cannabinoid-receptor antagonist. Rimonabant, a CB1 blocker once approved for weight management, was withdrawn because of psychiatric adverse effects. That makes the clean human experiment both ethically and practically difficult.
The endocannabinoid case therefore rests on convergence: repeated human blood changes, associations with mood, human evidence that opioids are not necessary for euphoria and anxiety reduction, and causal cannabinoid-receptor results for related behaviours in animals.
The full high is better understood as several systems meeting
Replacing “endorphins did it” with “anandamide did it” would preserve the same mistake. Runner’s high is not one laboratory outcome. It can include euphoria, anxiolysis, lower pain sensitivity, sedation, flow, altered time perception and increased motivation to continue.
The 2026 review gives endocannabinoids a central integrating role while describing contributions from other systems. Beta-endorphins appear more relevant to pain modulation and stress, especially during long or intense exercise. Dopamine helps shape motivation and reinforcement. Serotonin affects mood, while noradrenaline and adrenaline support arousal.
Brain-derived neurotrophic factor, or BDNF, supports plasticity rather than acting as an instant pleasure switch. Leptin links energy stores with motivation to move. The relative contribution of each system can shift with intensity, duration, metabolic state and individual biology.
A hard interval session can recruit stress and opioid systems while producing discomfort rather than euphoria. A steady moderate run may raise endocannabinoids and feel calmer. The route, setting, expectations and a runner’s familiarity with the effort can also influence the subjective result.
Not everyone experiences a recognisable high. The review notes that roughly 69 to 77 per cent of endurance runners in two studies said they had experienced one at least once. That retrospective measure depends on how participants define “high,” and it still leaves a substantial group who did not report the phenomenon.
What changed was the claim, not the existence of endorphins
ScienceBlog’s earlier account of the endocannabinoid evidence focused on the mouse receptor work and the 2021 human naltrexone trial. The newer review adds a useful correction to the correction: endocannabinoids are the leading mechanism, but the experience probably emerges from interaction rather than a molecular relay race with one winner.
The least defensible version of the old story is now clear. A rise in beta-endorphin measured in blood does not show that those circulating peptides entered the brain and caused euphoria. Blocking opioid receptors without erasing euphoria weakens the claim further.
The strongest current account places endocannabinoids near the centre. They have a plausible route across the barrier, usually rise after acute endurance exercise and are associated with core affective features. In mice, related benefits require cannabinoid receptors.
Important gaps remain. Direct cannabinoid blockade has not established the mechanism of human euphoria. Mouse behaviour cannot substitute for a subjective report. Blood measurements cannot reveal every signal occurring moment by moment inside a running brain.
Endorphins were not imaginary, and they did not become biologically irrelevant. What failed was the neat story that one circulating molecule explained everything. Runner’s high now looks less like a single switch and more like a temporary state assembled by several systems, with endocannabinoids conducting more of the music than scientists once realised.
This article reports research findings for general information and is not medical advice. Study results may not apply to an individual; consult a qualified health professional for personal guidance.