Benzodiazepines (alprazolam, clonazepam, diazepam, lorazepam)

Benzodiazepines (alprazolam, clonazepam, diazepam, lorazepam)

Benzodiazepines are the most reliable acute anti-anxiety drugs ever made. Within 15-60 minutes of taking one, the racing thoughts soften, the muscle tension drops, the panic backs off. They work by amplifying GABA, your brain's main braking neurotransmitter, so the whole system slows down. The four most commonly prescribed are alprazolam (Xanax), lorazepam (Ativan), clonazepam (Klonopin), and diazepam (Valium), and they differ mostly in how fast they hit and how long they last.
When you take them often the brain adapts to having its brakes pressed for you, and within a few weeks of daily use it cuts back its own GABA tone. That's tolerance, dependence, and a withdrawal syndrome that can be genuinely dangerous to stop cold. They're excellent tools for short, acute, high-acuity situations like a panic attack, an MRI, a flight you can't talk yourself out of, alcohol withdrawal under medical supervision. They're a bad choice for daily anxiety management, and a worse one combined with alcohol or opioids, where they're a major driver of overdose deaths. If you're using them, the goal is to use them rarely.

Deep-dive

Benzodiazepines bind to a specific site on the GABA-A receptor, a chloride channel made of five subunits with a binding pocket at the α+/γ− interface that's distinct from where GABA itself binds. They don't open the channel, they're positive allosteric modulators, meaning they make the channel more responsive when GABA is already present. The result is more chloride flowing into the neuron, which hyperpolarises it and makes it harder to fire. The cryo-EM structures resolved over the last few years confirmed the model that had been inferred from decades of electrophysiology, and identified a second, lower-affinity binding site in the transmembrane domain that contributes to the sedative and amnesic effects at higher doses.
Different α subunits in the GABA-A receptor mediate different effects. α1-containing receptors carry the sedation and the amnesia, α2 and α3 carry the anxiolytic effect, α5 is heavily expressed in the hippocampus and is the main driver of the memory impairment. Most marketed benzodiazepines are non-selective and hit all of them, which is why you can't really get the anxiolytic effect without some sedation and some memory cost. Newer α2/α3-selective compounds have been in development for years but none have made it through.
The four common ones. They share a mechanism but feel quite different in practice because of pharmacokinetics. The standard pharmacokinetic data puts the half-lives at roughly:
  • Alprazolam (Xanax): onset 15-30 minutes, half-life around 12 hours. Fast, short, no active metabolites. The fast on/off is what makes it the most reinforcing and the hardest to taper, since rebound anxiety can appear between doses
  • Lorazepam (Ativan): onset 15-30 minutes, half-life 10-20 hours. Cleared by glucuronide conjugation rather than CYP enzymes, so it has fewer drug interactions and is the standard pick in liver disease, ICU sedation, and alcohol withdrawal
  • Clonazepam (Klonopin): onset 30-60 minutes, half-life 18-50 hours. Long, smooth. Used for panic disorder, seizures, and sometimes as a once-daily anxiolytic. Less rebound between doses than alprazolam
  • Diazepam (Valium): onset under 15 minutes, parent drug half-life 20-100 hours, but it produces active metabolites including nordiazepam that extend the total exposure to 36-200 hours. Highly fat-soluble, hits fast, then accumulates with repeated dosing. The first-line choice for alcohol withdrawal and the drug everyone gets switched to before tapering off other benzos, because the long tail makes the come-off gentler
Approximate equivalencies, useful when reading the literature or comparing doses: 0.5 mg alprazolam = 1 mg lorazepam = 0.5 mg clonazepam = 10 mg diazepam.
What the evidence actually supports. Short-term, they work. Randomised trials in panic disorder consistently show large effects over 6-8 weeks. For generalised anxiety, social anxiety, and acute insomnia, they beat placebo over the same timeframe. The American Academy of Family Physicians summary is that benzodiazepines are effective for anxiety and sleep disorders at low doses for under 30 days, and that long-term use isn't supported by the evidence because tolerance develops to the anxiolytic and hypnotic effects faster than to the side effects. The 2025 Joint Clinical Practice Guideline on benzodiazepine tapering puts it bluntly: effective for a few weeks, other treatments are safer and equally or more effective longer-term, and almost everyone taking them daily for more than a month becomes physically dependent.
Tolerance and dependence. The mechanism is downregulation and uncoupling at the receptor. The brain detects chronic GABA-A potentiation and responds by reducing receptor density, altering subunit composition, and reducing endogenous inhibitory tone. Tolerance to the sedative and anticonvulsant effects appears within days to weeks, tolerance to the anxiolytic effect within weeks to months. Physical dependence is essentially universal after a few weeks of daily use, and is distinct from addiction, which (defined as compulsive use despite harm) happens in under 2% of people taking them as prescribed. You can be physically dependent and not addicted, this is normal, the body adapts to any chronic GABA-A potentiation.
Withdrawal. Stopping abruptly after sustained use produces a syndrome that includes rebound anxiety, insomnia, tremor, sweating, sensory hypersensitivity (sounds too loud, lights too bright), depersonalisation, and in serious cases seizures and psychotic episodes. The seizure risk is the part that makes benzodiazepine withdrawal medically dangerous, alongside alcohol it's one of the few withdrawal syndromes that can kill you. The classic review by Lader found dependence develops in roughly one-third of people on regular prescriptions of 4 weeks or longer, with higher dosage, more potent and shorter-acting agents (alprazolam in particular), and longer duration as the main risk factors. A subset of people develop protracted withdrawal symptoms that last months or years after the last dose, this is real but the exact prevalence is contested. Withdrawal management is the Ashton method: switch from short-acting to long-acting (usually diazepam), then taper by 5-10% every 1-2 weeks. Never stop cold turkey after sustained daily use.
Cognition. Acute doses impair attention, working memory, and especially episodic memory, this is consistent and dose-dependent. The harder question is what happens with chronic use. Two meta-analyses by Barker and colleagues found that long-term users showed broad cognitive impairment versus controls, with partial but not full recovery 6 months after withdrawal. An updated 2018 meta-analysis by Crowe and Stranks reached the same conclusion, residual impairment in most cognitive domains that persisted well past discontinuation. The dementia literature is messier. The 2014 BMJ case-control study by Billioti de Gage found a dose-dependent association between cumulative benzodiazepine exposure and Alzheimer's risk in over-65s, with long half-life agents (diazepam, clonazepam) carrying more risk than short half-life ones, and an odds ratio of 1.5-2.0 for sustained use. Subsequent studies have been mixed, some replicating the signal, others finding no effect once you adjust for the fact that anxiety and insomnia (the reasons people get prescribed benzos) are themselves prodromal symptoms of dementia. A 2024 Rotterdam Study population-based analysis found no overall association with dementia risk. The honest read is that there's a signal in older adults on long-term use, especially with long half-life agents, but the causality isn't settled. Either way, this is one more reason chronic daily use is the wrong move.
Falls and fractures in older adults. This one is unambiguous. A 2014 meta-analysis of 25 studies found a 25% increase in overall fracture risk and a 35% increase in hip fracture risk with benzodiazepine use, concentrated in those over 65. A 2017 meta-analysis found the risk is highest in the first two weeks after starting, when people are still adapting to the sedation and impaired balance. Z-drugs (zolpidem, zopiclone, eszopiclone) carry roughly the same risk, they're not safer.
The opioid and alcohol problem. This is the most important safety point on the page. Benzos alone are remarkably hard to overdose on, the GABA-A potentiation has a ceiling and doesn't push respiratory drive to zero. Combined with opioids or alcohol, that ceiling vanishes. In 2017, 21% of opioid overdose deaths in the US involved a benzodiazepine, and other estimates put the figure as high as 80% in some subsets of unintentional overdose deaths. The mechanism is synergistic respiratory depression at the brainstem, and it's worse with the more lipid-soluble agents (alprazolam, diazepam) that cross the blood-brain barrier fastest. Mixing benzos with opioids, alcohol, GHB, kratom at high doses, or any sedating substance is the single most reliable way to die from a benzodiazepine.
Women. The pharmacokinetic data on sex differences is mixed and clinically modest, but real. Women generally have higher exposure per dose because of lower body weight, higher body fat (relevant for the lipophilic agents that distribute into fat), and slower clearance of some agents. Oral contraceptives reduce benzodiazepine clearance, particularly for diazepam and the agents cleared by CYP3A4 (alprazolam, clonazepam, triazolam), so women on the pill end up with higher plasma levels and longer duration of effect. Lorazepam and oxazepam, which are cleared by glucuronide conjugation rather than CYP enzymes, are less affected by this. Women are also prescribed benzodiazepines substantially more often than men across nearly every country, which means the long-term consequences (dependence, cognitive effects, fall risk in older age) fall disproportionately on women without any matching clinical justification. The practical implications: women, especially women on the pill, can usually start at the lower end of the standard dose range and get the same effect, and the cognitive and sedation impacts may last longer per dose. In pregnancy, benzodiazepines cross the placenta, are associated with neonatal withdrawal and floppy infant syndrome when used near term, and the first-trimester teratogenicity signal (oral cleft) is weak but present, used in pregnancy only when the indication clearly outweighs the risk, like status epilepticus or severe alcohol withdrawal. Breastfeeding: short-acting agents at low doses (lorazepam, oxazepam) are generally compatible, the long-acting and lipophilic ones (diazepam, especially) accumulate in the infant.
Older adults. The combination of slower clearance, increased pharmacodynamic sensitivity, and more polypharmacy means benzos hit harder and stay longer. The triazolam pharmacokinetic study from Greenblatt showed AUC roughly doubled in elderly men compared with young men for the same dose, with comparable increases in sedation and cognitive impairment. The American Geriatrics Society Beers Criteria flags benzodiazepines as potentially inappropriate in over-65s for almost all indications. If they have to be used, glucuronidated agents (lorazepam, oxazepam, temazepam) at the lowest possible dose and shortest possible duration are the standard pick.
Paradoxical reactions. A small minority of users get the opposite of the expected effect, increased anxiety, agitation, disinhibition, aggression, sometimes hostility. More common in children, older adults, and people with personality disorders or pre-existing impulse control issues. If this happens, the drug doesn't work for you, switching agents rarely fixes it.

Dosage:

  • Alprazolam: 0.25-0.5 mg as needed for acute anxiety or panic, max 4 mg/day under medical supervision. Onset 15-30 min, lasts 4-6 hours subjectively. Avoid scheduled daily dosing if possible, the short half-life makes it the most reinforcing of the common benzos and the hardest to taper
  • Lorazepam: 0.5-1 mg as needed, max 6 mg/day. Onset 15-30 min, lasts 6-8 hours. The cleanest pharmacokinetic profile, preferred in older adults and anyone with liver issues or on multiple medications
  • Clonazepam: 0.25-0.5 mg, can be used once or twice daily for panic disorder under prescription. Onset 30-60 min, smooth and long-lasting. Less interdose rebound than alprazolam, generally easier to taper
  • Diazepam: 2-10 mg as needed, max 40 mg/day. Onset under 15 min, very long tail. The standard pick for tapering off other benzos, and for alcohol withdrawal under medical supervision
  • Use them episodically, not daily. The therapeutic window where they're net positive narrows fast with repeated use. A reasonable rule of thumb: if you're taking one more than 2-3 times a week for the same indication, you're on track for tolerance and dependence, and the underlying problem needs a different solution
  • Timing: Take on an empty stomach for fastest onset. With food, onset is delayed 30-60 minutes but peak levels are roughly the same
  • Women on oral contraceptives: start at the lower end of the dose range, especially for alprazolam, clonazepam, and diazepam. Lorazepam is less affected
  • Older adults (65+): halve the starting dose, prefer lorazepam, oxazepam, or temazepam, avoid diazepam and other long half-life agents unless there's a specific reason
  • Never stop abruptly after sustained daily use. Taper at 5-10% of the daily dose every 1-2 weeks, slower toward the end. Switching from a short-acting (alprazolam, lorazepam) to long-acting (diazepam) before tapering smooths the come-off

Here's what you can expect:

Within 15-60 minutes of an effective dose, the felt experience is the volume coming down on the anxious internal monologue, muscle tension easing, and a sense of being able to think clearly without the panic loop running underneath. Most people report feeling calm but not sedated at the lower end of the dose range. Higher doses or the more sedating agents (diazepam, longer-acting compounds) add drowsiness, mild slurring, and reduced coordination. Memory for events during the drug's peak action is often impaired, especially with alprazolam and lorazepam, this is the receptor pharmacology doing exactly what it does at the α5 site.
Acute panic, anticipatory anxiety before a known stressor, alcohol withdrawal, status epilepticus, acute mania while waiting for other medications to take effect, this is where benzos are excellent.
With chronic daily use, the felt effect erodes within weeks to a few months as tolerance develops. The anxiety reliably comes back, often worse than baseline once rebound and withdrawal start showing through between doses, and people end up taking more for diminishing returns. This is the pattern that leads to long-term dependence, and the trajectory is hard to reverse without a structured taper.

Side effects & risks:

  • Sedation, drowsiness, slowed reaction time, impaired coordination. Dose-dependent, more pronounced with the long half-life agents and in older adults. Don't drive within 8 hours of an effective dose
  • Cognitive impairment. Acute memory and attention impairment is reliable and dose-dependent. Long-term daily use is associated with persistent cognitive deficits that only partially recover after discontinuation
  • Tolerance and physical dependence are essentially universal after 4+ weeks of daily use. Not the same as addiction, but it means you can't stop abruptly without consequences
  • Withdrawal syndrome after sustained use includes rebound anxiety, insomnia, tremor, sensory hypersensitivity, and at higher doses or with abrupt cessation, seizures. Benzodiazepine withdrawal is one of the few withdrawal syndromes that can be fatal. Always taper under medical supervision
  • Respiratory depression with opioids or alcohol is the most serious acute risk and accounts for the majority of benzodiazepine-related deaths. Don't combine. This includes prescription opioids, methadone, buprenorphine, and high-dose kratom
  • Falls and fractures in older adults, with a 25-35% increase in risk, highest in the first two weeks after starting
  • Disinhibition and paradoxical agitation in a small minority, more common in children, older adults, and those with personality or impulse control disorders
  • Pregnancy: crosses the placenta, neonatal withdrawal and floppy infant syndrome with use near term, weak first-trimester teratogenicity signal. Use only when the indication clearly outweighs the risk
  • Driving and machine operation impairment can last 12+ hours after a single dose of long-acting agents, longer with chronic use
  • Sleep architecture: suppresses deep sleep (slow wave sleep) and REM. You sleep through the night but the sleep is lower quality, which is part of why benzos don't fix the underlying problem in chronic insomnia
  • Older adults: higher plasma levels per dose, slower clearance, increased pharmacodynamic sensitivity, more falls, more cognitive impairment, possible dementia signal with chronic use of long half-life agents. Use cautiously and briefly if at all
🩸

Blood markers

Liver enzymes (ALT, AST, GGT), baseline if you anticipate using benzodiazepines regularly or are on other hepatically cleared medications. Most benzodiazepines are processed through CYP3A4 (alprazolam, clonazepam, diazepam) and impaired liver function meaningfully slows clearance, with diazepam half-life in particular extending to several days in moderate cirrhosis. Lorazepam, oxazepam, and temazepam are cleared by glucuronide conjugation and are largely spared.
Creatinine and eGFR, baseline. Renal clearance matters less than hepatic for most benzodiazepines, but severe kidney impairment slows elimination of active metabolites and is worth knowing.
Complete blood count and basic metabolic panel as a general baseline, especially before any extended course.
For older adults specifically: baseline cognitive screening (something simple like the Mini-Cog or MoCA) before any sustained course, with a recheck at 3-6 months. Subtle cognitive decline is the most underdiagnosed long-term effect.
For most people using benzodiazepines situationally (a handful of times a year for acute events, MRIs, flights), no specific bloodwork is needed. The people who actually need baseline testing are those on or considering daily/weekly use, anyone with liver or kidney disease, older adults, and anyone combining with other sedating medications. The most important monitoring isn't a blood test, it's an honest log of how often you're taking it and whether the dose is creeping up.
Prescription-only in nearly all countries and one of the most tightly regulated drug classes.