IUPAC: 7-chloro-1-methyl-5-phenyl-1,3-dihydro-2H-1,4-benzodiazepin-2-one · C₁₆H₁₃ClN₂O · MW 284.75 g/mol · CAS 439-14-5 · ATC N05BA01
Diazepam (Valium). The archetypal 1,4-benzodiazepine — Leo Sternbach's 1963 successor to chlordiazepoxide, the best-selling drug in the United States from 1969 to 1982. A positive allosteric modulator of the GABAA receptor at the benzodiazepine site. Anxiolytic · anticonvulsant · muscle relaxant · sedative-hypnotic. Trade/context names: Valium, Diastat, Valtoco, "vallies," "blues."
Diazepam does not open the chloride channel by itself. It is a positive allosteric modulator (PAM) that binds an extracellular pocket at the α+/γ2− subunit interface of the pentameric GABAA receptor — the "benzodiazepine site," structurally homologous to the two orthosteric GABA sites that sit at the β+/α− interfaces. When GABA is present, diazepam binding raises the receptor's affinity for GABA and increases the frequency of channel openings. This is the mechanistic contrast that matters clinically: benzodiazepines increase opening frequency; barbiturates increase opening duration and, at high dose, gate the channel directly. That single difference is why isolated benzodiazepine overdose has a ceiling and barbiturate overdose does not — and why the ceiling evaporates the moment a second CNS depressant is added (see §04).
GABAA is a ligand-gated anion channel. GABA binding at the two β+/α− sites opens the pore; Cl⁻ (and HCO₃⁻) flows down its electrochemical gradient, hyperpolarizing the neuron and raising firing threshold. Diazepam adds nothing without GABA — it is pure gain, not a switch.
The conserved α-subunit His-101 (α1 numbering) is required for high-affinity diazepam binding. α4 and α6 carry an arginine at this position instead — making α4- and α6-containing receptors "diazepam-insensitive" (Ki in the µM range). Point-mutant (H101R) mouse lines dissected each subtype's behavioral role.
α1βγ2 receptors (~60% of brain GABAA) mediate sedation, anterograde amnesia, and anticonvulsant action and much of the reinforcing/abuse-related effect. This is the subtype the "z-drugs" (zolpidem) target preferentially.
α2- and α3-containing receptors (limbic system, spinal cord) carry the anxiolytic and myorelaxant effects. Diazepam is non-selective — it hits α1, α2, α3, and α5 with comparable affinity, which is exactly why it is simultaneously anxiolytic, sedating, amnestic, and muscle-relaxing.
α5βγ2 receptors concentrate in the hippocampus and shape learning and memory; diazepam's α5 potentiation contributes to its amnestic profile. α5-selective negative modulators are studied as pro-cognitive agents — the mirror image of this pharmacology.
Diazepam also binds the 18-kDa translocator protein (TSPO) — the historic "peripheral benzodiazepine receptor" on outer mitochondrial membranes — at lower affinity. TSPO governs cholesterol import for steroidogenesis and is not the source of anxiolysis, but it is why diazepam is not a "clean" single-target ligand.
Because the effect is use-dependent on ambient GABA, benzodiazepines have a self-limiting envelope in isolation: they can only amplify inhibition where GABA is already being released. Barbiturates, ethanol at high concentration, and — critically — the additive load of an opioid on the same brainstem respiratory circuitry, remove that ceiling.
Diazepam is highly lipophilic — it is absorbed rapidly and near-completely after oral dosing (bioavailability ~93–100%), crosses the blood–brain barrier within minutes, and then redistributes into fat, which shortens the subjective effect of a single dose while the drug lingers systemically for days. The defining feature of diazepam PK is not the parent compound's already-long half-life (~20–70 h) but its cascade of pharmacologically active, even longer-lived metabolites. Steady-state accumulation over repeated dosing is substantial, and clearance is markedly prolonged in the elderly and in hepatic impairment.
Metabolism cascade: Diazepam is N-demethylated by CYP2C19 to nordazepam (desmethyldiazepam) and 3-hydroxylated by CYP3A4 to temazepam. Both routes converge on oxazepam, which is finally glucuronidated by UGT enzymes and cleared renally. Every intermediate except the glucuronide is active at the BZD site — so a single diazepam dose is really a slow-release cocktail of four benzodiazepines.
Nordazepam (marked ★) is the pharmacological workhorse of chronic dosing: with a half-life of 40–100 hours (longer in older adults), it accumulates to concentrations exceeding the parent drug and drives the sustained, "smooth" anxiolysis for which diazepam is prescribed — as well as the long tail of impairment and the difficulty of ever fully clearing the drug. Temazepam and oxazepam are marketed as benzodiazepines in their own right.
CYP2C19 poor metabolizers (~2–5% of Europeans, ~15–20% of East Asians) clear diazepam far more slowly and reach higher steady-state levels. CYP3A4 inhibitors — grapefruit juice, azole antifungals, macrolides, ritonavir — raise exposure further. Because both parent and metabolites are so long-lived, dose changes take a week or more to reach a new steady state, and stacking doses "because it isn't working yet" is a common route to unintentional over-sedation.
The subjective experience of diazepam is dominated by anxiolysis, muscle relaxation, sedation, and anterograde amnesia, with a mild euphoria and disinhibition that underwrites its recreational and reinforcing value (largely α1-mediated). But the clinically decisive story is adaptation: the GABAergic system does not tolerate sustained potentiation without remodeling itself, and that remodeling is what produces tolerance, dependence, and a withdrawal syndrome that can kill.
With chronic exposure, GABAA receptors undergo subunit trafficking and uncoupling of the BZD site from the GABA site. Sedative and anticonvulsant tolerance develop fastest (days to weeks); anxiolytic tolerance is slower and partial. The functional result: the same dose delivers progressively less effect, inviting dose escalation.
To offset chronic inhibition, the CNS down-regulates GABAergic tone and up-regulates glutamatergic (NMDA) drive. As long as diazepam is present, the system is balanced. Remove it and the compensations are unmasked — the brain is now wired for hyperexcitability. Dependence can develop within weeks of daily therapeutic use; it is a physiological adaptation, not a character flaw.
Abrupt cessation after sustained use produces rebound anxiety and insomnia, tremor, sweating, perceptual disturbance, and — in severe cases — generalized tonic-clonic seizures, psychosis, and delirium, mechanistically analogous to alcohol withdrawal. A subset of users experience protracted withdrawal lasting months. Because of diazepam's long half-life, onset can be delayed a week or more after the last dose, masking causation.
The evidence-based response is a gradual dose taper, often over months, sometimes after converting shorter-acting benzodiazepines to long-acting diazepam precisely because its slow decline smooths the withdrawal gradient (the "Ashton manual" approach). Never stop abruptly after regular use. Taper is a medical procedure, not a test of willpower.
This is the central asymmetry of diazepam: in isolation it is one of the safer psychoactive drugs ever made — the therapeutic-index margin against fatal respiratory depression is wide — yet it creates a dependence that makes stopping genuinely hazardous, and it becomes lethal the instant it is stacked with another respiratory depressant. The molecule is not the danger. The combination and the discontinuation are.
Evidence-based, non-judgmental. Diazepam's risks are dominated by two things: what you combine it with, and how you stop it. Both are manageable with accurate information.
| Target (subtype) | Affinity | Rel. | Action |
|---|---|---|---|
|
α1β3γ2
GABAA · sedation / anticonvulsant
|
Ki ≈ 14 nM
human recomb.
|
PAM | |
|
α5β3γ2
GABAA · hippocampal / memory
|
Ki ≈ 11 nM
|
PAM | |
|
α3β3γ2
GABAA · anxiolytic / myorelaxant
|
Ki ≈ 15 nM
|
PAM | |
|
α2β3γ2
GABAA · anxiolytic
|
Ki ≈ 20 nM
|
PAM | |
|
α6β3γ2
GABAA · "diazepam-insensitive"
|
Ki ≈ 3,000 nM
Arg at His-101
|
≈ inactive | |
|
TSPO
18-kDa translocator (periph. BZD)
|
µM-range
secondary site
|
Binder |