IUPAC: ethanol · CH₃CH₂OH · C₂H₆O · MW 46.07 g/mol · CAS 64-17-5 · ATC V03AB16 / D08AX08 · ChEMBL CHEMBL545 · the oldest and most-consumed psychoactive drug on Earth
Ethanol (ethyl alcohol, "alcohol," "booze"). A two-carbon amphipathic solvent that happens to be psychoactive. It is not a lock-and-key ligand: too small, too weak, and too promiscuous for a single high-affinity pocket. Instead it perturbs several ion channels at once at millimolar concentrations — a positive allosteric modulator of GABAA receptors and a non-competitive NMDA-receptor antagonist, with additional actions on glycine receptors, neuronal nicotinic receptors, and GIRK/Kir3 potassium channels. That dual "brake-on, accelerator-off" profile — enhancing inhibition while suppressing excitation — is the whole story: the intoxication, the tolerance, and a withdrawal syndrome that can kill.
The single most important fact about ethanol pharmacology is the concentration scale. A cocaine or a fentanyl molecule works at nanomolar occupancy — one drug molecule per receptor, held by a shaped pocket. Ethanol has no such pocket. Behaviorally relevant blood concentrations run from roughly 10 mM (mild intoxication) to 100 mM (coma / respiratory arrest) — six orders of magnitude weaker than a classical drug. At those concentrations there are literally billions of ethanol molecules per neuron, transiently occupying shallow, low-affinity, water-accessible cavities at protein–protein and protein–lipid interfaces of several membrane proteins at once. Ethanol is not selective because it cannot be selective. Its pharmacology is the summed perturbation of many weakly coupled targets.
Two of those targets dominate the clinical picture and pull in opposite directions on the glutamate/GABA balance that sets cortical excitability: it potentiates GABAA (the brain's main inhibitory chloride channel) and it blocks the NMDA receptor (the main excitatory, calcium-permeable glutamate channel). Turn inhibition up and excitation down simultaneously and you get the classic depressant arc — anxiolysis, disinhibition, sedation, ataxia, amnesia, then anaesthesia and death — plus the exact adaptations that make stopping dangerous.
Ethanol enhances GABA-gated Cl⁻ current, deepening inhibition. The clearest, low-millimolar sensitivity is at extrasynaptic α4/α6-βδ receptors mediating tonic inhibition (Wallner/Hanchar/Olsen, PNAS 2003) — though this low-dose effect is genuinely contested (Borghese et al.). Synaptic γ2 receptors respond at higher (≳ 40–100 mM) concentrations. Ethanol shares residues in the TM2/TM3 cavity with volatile anaesthetics (Mihic et al., Nature 1997).
Ethanol non-competitively inhibits NMDA-activated current (Lovinger, White & Weight, Science 1989) — ~50% block near 50 mM, dose-dependent from ~5 mM. It does not occupy the glutamate or glycine sites; it acts at hydrophobic residues near the TM3/TM4 channel domain. GluN2B-containing receptors are the most ethanol-sensitive. This is the excitatory arm of intoxication and the seed of withdrawal.
The other inhibitory Cl⁻ channel: ethanol potentiates α1 glycine receptors at 10–100 mM, contributing to brainstem and spinal-cord depression (motor incoordination, respiratory drive). The Mihic 1997 alcohol-cavity residues sit in the same TM2/TM3 region as GABAA — one structural motif, two inhibitory channels.
Ethanol directly activates G-protein-gated inwardly-rectifying K⁺ channels (GIRK2/Kir3.2) at behaviorally relevant concentrations, hyperpolarizing neurons independently of GABA (Kobayashi et al., Lewohl et al., Nat. Neurosci. 1999). A hydrophobic pocket in the cytoplasmic domain was later resolved — one of the few actual "alcohol pockets" known.
Ethanol modulates neuronal nicotinic acetylcholine receptors (subtype-dependent potentiation/inhibition) and potentiates 5-HT₃ receptors — both feed dopaminergic reward signalling in the VTA/nucleus accumbens, part of why ethanol is reinforcing despite being a depressant.
Ethanol raises mesolimbic dopamine indirectly and triggers endogenous opioid (β-endorphin) release; this is the rationale for naltrexone in alcohol-use disorder — blunting the µ-opioid-mediated reward blunts craving. The reward circuitry is downstream of the ion-channel actions, not a direct ethanol receptor.
Because every one of these actions is low-affinity and reversible, ethanol's effect tracks its concentration almost in real time — there is no reservoir of tightly bound drug. That is why intoxication rises and falls with the blood-alcohol curve, and why the brain, chronically bathed in a drug it cannot escape, remodels its glutamate and GABA machinery to compensate (see §03).
Ethanol is small, water-soluble, and freely miscible; it needs no transporter and distributes into total body water (Vd ≈ 0.5–0.6 L/kg), which is why body composition and sex change the blood level for the same dose. Absorption is rapid from stomach and (mainly) proximal small intestine; food slows gastric emptying and blunts the peak. Protein binding is negligible. The defining feature is elimination: ~90–98% of a dose is metabolized in the liver, with only 2–10% excreted unchanged in breath, urine, and sweat (the basis of the breathalyzer).
The metabolic enzymes saturate at very low blood concentrations. Hepatic alcohol dehydrogenase (ADH) has a low Km (~1 mM), so at any intoxicating level it is running flat-out. The consequence is zero-order (Michaelis–Menten-saturated) kinetics: ethanol is cleared at a roughly constant amount per unit time, not a constant fraction. Typical adult clearance is ~0.015–0.02 g/dL per hour (≈ 7–10 g, roughly one standard drink, per hour). Double the dose and you double the time to sober, not the rate — there is no way to speed it up, and "sobering" tricks do nothing.
The oxidation cascade: ADH oxidizes ethanol to acetaldehyde (using NAD⁺ → NADH). Acetaldehyde is a reactive, protein- and DNA-adducting Group 1 carcinogen — the actual toxic and carcinogenic principle of drinking. Mitochondrial aldehyde dehydrogenase-2 (ALDH2) then rapidly converts it to harmless acetate, cleared as acetyl-CoA → CO₂ + H₂O. A second, inducible route — microsomal CYP2E1 (the MEOS) — becomes significant at high concentrations and in chronic drinkers, contributing to metabolic tolerance and to oxidative stress.
The ALDH2 bottleneck is where genetics and drugs act. Roughly 540 million East Asians carry the loss-of-function ALDH2*2 (Glu504Lys) allele: acetaldehyde accumulates after a single drink, producing the "Asian flush" — facial flushing, tachycardia, nausea. It is protective against alcoholism and simultaneously raises esophageal-cancer risk from the same acetaldehyde load. Disulfiram (Antabuse) is a drug that deliberately inhibits ALDH2: drink on it and acetaldehyde spikes, causing a violently aversive reaction. Metronidazole, some cephalosporins, and coprine (inky-cap mushrooms) do the same accidentally.
The NAD⁺ → NADH shift from ADH/ALDH is itself a metabolic drug effect: the reduced redox state drives lactic acidosis, hypoglycaemia (blocked gluconeogenesis — dangerous in fasting or diabetic drinkers), hyperuricaemia (gout), and fatty liver (steatosis). Ethanol's harm is not only neuronal; it hijacks the cell's central redox currency.
Ethanol's acute effect is net inhibition. Sustain that for weeks and the brain fights back: it up-regulates NMDA/glutamatergic drive and down-regulates GABAA tone to restore normal excitability while the drug is present. As long as ethanol is on board, the system is balanced. Remove it, and the compensations are unmasked — a brain now wired for hyperexcitability, flooded with unopposed glutamate. This is the mechanistic mirror image of benzodiazepine withdrawal (§ Diazepam), and it is why the two are cross-tolerant and why one treats the other.
Three layers stack: metabolic (CYP2E1 induction speeds clearance), pharmacodynamic (NMDA up-regulation, GABAA subunit remodeling and uncoupling), and behavioral/learned tolerance. The same blood level delivers progressively less effect, driving escalation — while the lethal-dose ceiling barely moves, so the therapeutic window narrows.
Repeated cycles of intoxication and withdrawal progressively lower the seizure threshold — an electrophysiological "kindling" of limbic circuits. Someone on their fifth detox is at far higher risk of seizures and delirium tremens than a first-timer at the same intake. Every unmanaged withdrawal sensitizes the brain for the next one. This is a core argument against repeated abrupt "cold-turkey" attempts.
6–24 h: tremor, anxiety, sweating, tachycardia, insomnia. 12–48 h: generalized tonic-clonic withdrawal seizures; alcoholic hallucinosis. 48–96 h: delirium tremens — clouded consciousness, disorientation, hallucinations, and autonomic storm (fever, hypertension, tachycardia). Untreated DTs historically killed ~15–35%; even treated, mortality is ~1–5%. Unlike opioid withdrawal, alcohol withdrawal is directly lethal.
Benzodiazepines are first-line (long-acting chlordiazepoxide or diazepam; lorazepam/oxazepam in liver failure): being GABAA PAMs, they are cross-tolerant with ethanol and restore inhibitory tone while the brain re-equilibrates. Symptom-triggered dosing to a CIWA-Ar scale outperforms fixed schedules. Phenobarbital is an increasingly used alternative/adjunct. Parenteral thiamine before glucose prevents Wernicke encephalopathy. Managed withdrawal is a medical procedure — not willpower.
This is the central asymmetry: acute ethanol has a comparatively wide single-drug therapeutic window (lethal BAC ~0.4–0.5%, ~5× the legal driving limit), yet chronic use builds a dependence in which stopping is a medical emergency, and the drug becomes acutely lethal the instant it is stacked with another respiratory depressant. The molecule is weak. The adaptation and the combinations are what kill.
No moralizing, just the pharmacology of staying alive. Ethanol is legal, normalized, and — by aggregate harm — the most damaging recreational drug in wide use. In the Nutt et al. multicriteria analysis (Lancet 2010), summing harm-to-user and harm-to-others, alcohol scored highest (72/100), above heroin (55) and crack (54). That ranking is driven mostly by harm to others (violence, collisions, fetal harm, social cost) plus organ toxicity — not acute overdose. The risks below are the ones you can actually manage.
| Target | Potency | Rel. | Action |
|---|---|---|---|
|
GABAA α4/6βδ
extrasynaptic · tonic inhibition
|
EC ~1–10 mM
low-dose; contested
|
PAM | |
|
GIRK2 / Kir3.2
G-protein-gated K⁺ channel
|
act. ~10–40 mM
|
Activator | |
|
Glycine R (α1)
inhibitory Cl⁻ channel
|
EC ~10–100 mM
|
PAM | |
|
NMDA (GluN2B)
excitatory glutamate channel
|
IC₅₀ ~30–60 mM
non-competitive
|
Antagonist | |
|
GABAA γ2 (synaptic)
phasic inhibition
|
EC ≳ 40–100 mM
|
weak PAM | |
|
5-HT₃ / nAChR
ligand-gated cation channels
|
~mM
subtype-dependent
|
Modulator |