The benzodiazepine taper, in four focused lessons. Yours free.

Why this class needs a careful hand

Benzodiazepines and Z-drugs (the “Z-drugs” being zopiclone, zolpidem, and zaleplon) are grouped together for a reason. They are all positive allosteric modulators of the GABA-A receptor, meaning they boost the effect of GABA, the brain’s main inhibitory (“calming”) signal (p.313). Because they share that mechanism, they share a withdrawal picture. Stopping either class, especially quickly, can be difficult and occasionally severe.

Physical dependence here is old news, not a surprise. It has been recognized for nearly six decades, and in 2020 the FDA updated the boxed warning on all marketed benzodiazepines to re-emphasize that withdrawal syndromes “could persist for more than 12 months after stopping” (p.309). The same FDA review found that withdrawal was often misread as a return of the original illness, which led to higher doses and tapers that were far too fast (p.309).

Dependence is not addiction

This distinction carries clinical weight. As the FDA put it, “physical dependence is not synonymous with addiction; a patient may be physically dependent on a drug without having an addiction to it” (p.311). Tolerance, physical dependence, and withdrawal are expected biological consequences of chronic treatment with these drugs. They do not, on their own, signal addiction.

The numbers make the point. Only 2% of people who use benzodiazepines have a use disorder, while 17.1% have misused them once or more, and the majority taking them as prescribed will become physically dependent simply through adaptation (p.312). Mislabeling that dependence as addiction has consequences. A request to manage withdrawal can be dismissed as “drug seeking” and refused, and because the syndrome can include seizures and can be life-threatening, that refusal can be catastrophic (p.312). This course is about people using these drugs as prescribed; those with addiction need separate support that is outside its scope (p.312).

What withdrawal looks like, and how severe

The syndrome is broad because these drugs touch many systems. The Maudsley groups the effects into physical and psychological categories (p.316–317). Physical symptoms include sweating, headache, tremor, muscle stiffness and spasm, and a striking range of sensory disturbances: tinnitus, blurred vision, hypersensitivity to light and sound, numbness, and formication, the sensation of insects crawling under the skin (p.316). Perceptual disturbances, such as the floor seeming to undulate, can occur. Psychological symptoms include anxiety, panic, depression, irritability, and, in abrupt or very rapid withdrawal, delirium and psychotic features (p.316–317).

Two points matter for severity. First, the symptoms can be life-threatening, with fatalities recognized; seizures, delirium, and catatonia can occur, especially when the drug is stopped abruptly, and suicidality is a recognized withdrawal effect that can appear even in people who never had it before (p.309). Second, the familiar symptoms are easy to misread. Because this class is so often prescribed for anxiety and insomnia, and because anxiety and insomnia are also common withdrawal symptoms, the two are easily confused (p.316). One distressing presentation, akathisia, a state of severe agitation, restlessness, and inner terror, is frequently mistaken for a manic state, an anxiety or panic disorder, or even ADHD (p.318).

How half-life shapes the timeline

Onset tracks the drug’s pharmacokinetics. Withdrawal effects “often” begin a day or two after stopping for shorter-acting benzodiazepines, and several days for longer-acting drugs, though delayed onset is also recognized (p.309). The time to peak severity varies, and symptoms can fluctuate over time (p.309).

It helps to picture the agents on a spectrum of half-life. Shorter-acting drugs such as lorazepam, alprazolam, and temazepam clear faster, which produces a sharper drop in drug level and is associated with more severe acute withdrawal (p.309). Longer-acting drugs such as diazepam, clonazepam, and chlordiazepoxide clear more slowly, which tends to soften onset, though both short- and long-acting drugs can cause protracted syndromes (p.309). That difference in half-life is exactly why the next lessons build the taper around a switch to diazepam: a long, smooth tail makes the dose reductions gentler.

When withdrawal lingers

For most people, acute symptoms wane over weeks. For a minority they do not. The Maudsley estimates that up to 15% of people who stop long-term benzodiazepines will experience a protracted withdrawal syndrome lasting months and sometimes years (p.309), and elsewhere puts the protracted figure at roughly 10% to 25% of long-term users (p.320). The FDA describes protracted withdrawal as persisting beyond 4 to 6 weeks, with symptoms in some patients lasting “years,” and highlights anxiety, cognitive impairment, depression, insomnia, formication, motor symptoms, paraesthesia, and tinnitus (p.320).

This course is built on the opposite of the FDA’s cautionary finding. Because shorter-acting drugs and faster tapers are linked to greater withdrawal difficulty (p.318), the safest path is a slow, individualized reduction, the subject of the lessons that follow.

Source: Horowitz MA, Taylor D. The Maudsley Deprescribing Guidelines (Wiley Blackwell, 2024), Ch.3 — “Safe Deprescribing of Benzodiazepines and Z-drugs” (p.309–321).

Why switch at all

Short-acting benzodiazepines clear the body quickly. Between doses, blood levels fall, and a sensitized patient can feel withdrawal creeping in hours before the next tablet is due. That is inter-dose withdrawal, and it makes a taper miserable before it has even started.

A drug with a long half-life behaves differently. Levels stay flatter through the day, so the body experiences a more gradual change as the dose comes down. The Maudsley puts the theory plainly: a longer-acting drug “should cause a more gradual change in plasma levels following reductions that would minimise withdrawal symptoms,” and this “may be particularly helpful for patients who are experiencing inter-dose withdrawal from shorter-acting benzodiazepines” (p.335).

Diazepam is the usual choice. Its half-life is long, roughly 24 to 48 hours and up to 100 hours counting the active metabolite, and it comes in small-dose tablets and liquids that make fine reductions practical later on (p.335–336).

What the evidence actually says

This is worth stating honestly. The Maudsley notes “mixed views” on switching, and a Cochrane review “did not find strong evidence one way or the other” (p.335). So the switch is not proven to help. It is a reasonable, theory-backed option for the right patient, not a step every taper needs. Many people taper successfully from their original drug and never switch at all.

Reading the equivalence table

Maudsley Table 3.7 lists approximate diazepam equivalents. Diazepam 10 mg is the anchor. A few common conversions:

• Lorazepam 1 mg ≈ diazepam 10 mg
• Alprazolam 0.5 mg ≈ diazepam 10 mg
• Temazepam 20 mg ≈ diazepam 10 mg
• Clonazepam 0.5 mg ≈ diazepam 10 mg

Notice how potent the short-acting agents are. Half a milligram of alprazolam carries the same equivalent weight as ten milligrams of diazepam, which is exactly why abrupt swaps go wrong.

The table comes with a warning the authors print in bold terms: these equivalencies “are not based on careful studies but based on expert opinion, and so can only be seen as rough estimates, and there can be significant inter-individual differences” (p.335). Treat the number as a starting hypothesis, not a fixed exchange rate.

How to do the switch

Because the equivalences are approximate, the switch is performed gradually and partially rather than all at once. The Maudsley is direct: “switching should be performed step-wise rather than abruptly to allow a period of accommodation to the new drug and to check for withdrawal symptoms” (p.335). A workable shape looks like this:

1. Convert a portion, not the whole dose. Replace one part of the short-acting benzodiazepine with its diazepam equivalent and leave the rest in place.
2. Watch the patient, not just the math. Patient-reported symptoms lead. Adjust the amount swapped based on how the person feels, so as to minimise withdrawal (p.335).
3. Repeat in steps until the patient is on diazepam alone.
4. Stabilize before tapering. Allow a settling period, “e.g. 2 weeks or longer,” on the new drug before the dose reduction begins (p.335).

One practical nuance: because sensitized patients can react to even small swings in plasma level, some need to take their longer-acting benzodiazepine more than once a day, even though the long half-life would suggest once-daily dosing is enough (p.335).

When NOT to switch

The switch is not always the kinder path.

• Diazepam can accumulate. Its long half-life, including the active metabolite, means levels build up over time. Extra precautions apply in hepatic dysfunction, where diazepam and other long-acting drugs may accumulate further (p.335).
• The familiar drug is sometimes easier. Some patients “find switching difficult or prefer to taper from the original, familiar drug” (p.335). Comfort and confidence matter in a taper that may run for months.
• Z-drugs may not need it. Z-drugs with short half-lives are largely eliminated between doses, so they theoretically cause less adaptation. Direct tapering from these may be more sensible, with a switch reserved mainly for patients who hit inter-dose withdrawal or who struggle to taper directly (p.335).

The principle to carry forward: switching to diazepam is one technique for smoothing a taper, useful chiefly for inter-dose withdrawal and difficult direct tapers. It is approximate, individualized, and optional. The hyperbolic, symptom-guided reduction that follows the switch is the subject of the next lessons.

Source: Horowitz MA, Taylor D. The Maudsley Deprescribing Guidelines (Wiley Blackwell, 2024), Ch.3 — switching to longer-acting benzodiazepines (p.335) and dose-equivalence Table 3.7 (p.336).

Why fixed milligram steps stop working

The relationship between a benzodiazepine dose and its effect at the GABA-A receptor is hyperbolic, not linear (p.332). When little drug is present, most receptors are empty, so even small dose changes produce large changes in effect. As the dose climbs, receptors saturate and each added milligram does less (p.332).

That shape has a direct clinical consequence. Cutting by a fixed amount, say 15 mg of diazepam each step, produces an increasingly large reduction in receptor occupancy as the dose falls. The Maudsley illustrates it: a 5 mg drop from 50 mg lowers GABA-A occupancy by about 1.5 percentage points, but the same 5 mg drop from 10 mg lowers it by about 11.7 points, roughly eight times the change in effect (p.332–333). This matches what patients report. The same-sized reduction causes greater withdrawal at lower doses (p.332).

Reduce by a proportion, aim for equal effect

To produce equal-sized reductions in receptor occupancy at each step, the dose reductions themselves have to get smaller and smaller as the total comes down. That pattern is called hyperbolic tapering because it follows the curve (p.333).

Table 3.6 in the Maudsley puts numbers to this. To lower GABA-A occupancy by 5 percentage points each time, the dose reduction near 60 mg is about 11.7 mg, but near the bottom of the taper it shrinks to roughly 1.3–1.4 mg (p.333). Equal reductions in occupancy are likely to produce an equivalent intensity of withdrawal, which is why the Maudsley’s suggested regimens reduce diazepam by about 4 mg when a patient is on 60 mg a day, but by only about 0.5 mg once the daily dose is down near 2.5 mg (p.334). Same felt effect, very different milligram steps.

This is the load-bearing idea of the lesson: pace the taper to the size of the effect, not the size of the number.

Hold to stabilize between steps

Hyperbolic tapering needs many intermediate steps, with holds built in. The drug-specific regimens in the Maudsley make reductions every 1 to 4 weeks, sometimes described as “cut and hold,” which lets symptoms settle before the next cut (p.342). An alternative is micro-tapering, where a small change is made each day rather than in discrete steps; the Maudsley notes a micro-taper rate can be derived by dividing a step from the regimens across 7, 14, or 28 days (p.342).

Independently, peer-led withdrawal communities have arrived at a similar place. Many recommend reducing by a proportion of the most recent dose each period, often about 10% (calculated on the last dose, so the steps become progressively smaller), with reductions every 2 to 4 weeks (p.334). The convergence of receptor-occupancy math and lived patient experience is reassuring, though the Maudsley is clear there is considerable individual variation (p.334).

Making up the small doses

Because the final steps are tiny, often less than 1 mg diazepam-equivalent, patients need a way to make small doses (p.334). The Maudsley describes several practical methods (p.338–341):

• Manufacturers’ liquids and dilutions. Diazepam, lorazepam, and clonazepam come in liquid forms. A manufacturer’s example dilutes 0.5 mL of a 5 mg/mL diazepam solution into 4.5 mL of water (a 10-fold dilution) to make 0.5 mg/mL, from which 0.25 mg is 0.5 mL. Because dilution stability is not guaranteed, the dose is taken soon after preparation and the remainder discarded (p.339–340).
• Tablet splitting. Tablet cutters divide scored tablets more accurately than hand-splitting; useful for early steps, but very sensitized patients may need a more precise method (p.339).
• Dispersing tablets in water. Many non-coated, immediate-release benzodiazepine tablets disperse in water to make a suspension that is shaken well and measured by volume (p.340).
• Compounded preparations. Compounding pharmacies can prepare smaller-dose tablets, capsules, or “tapering strips,” from a reputable pharmacy (p.340).

The Maudsley also advises against tapering by skipping doses for benzodiazepines other than long-acting ones like diazepam, because most have half-lives of 24 hours or less and every-other-day dosing lets plasma levels fall far enough to trigger inter-dose withdrawal (p.338–339).

Slow is fastest

Benzodiazepine tapers can run for weeks to many months, and when symptoms get severe the Maudsley’s preference is to slow the taper rather than add another medication to mask it (p.342). The program should stay flexible and adjustable so the process is tolerable, with close monitoring of symptoms and access to the formulations needed for small reductions (p.342). The patient sets the pace, and going slower now is usually what gets the taper finished.

Source: Horowitz MA, Taylor D. The Maudsley Deprescribing Guidelines (Wiley Blackwell, 2024), Ch.3 — hyperbolic tapering (p.332–334), Table 3.6 (p.333), small-dose formulations Table 3.8 (p.338–341), micro-tapering (p.342).

What the Z-drugs actually are

Zopiclone, zolpidem, and zaleplon are often described to patients as “not benzodiazepines,” and that is technically true. The Maudsley calls them non-benzodiazepine hypnotics. But pharmacologically they sit very close. Each acts “in a similar manner to benzodiazepines as a positive allosteric modulator of the GABA-A receptor” (p.480, 485, 490). Zolpidem, for instance, binds the omega-1 subtype benzodiazepine site (p.485). The practical takeaway is that everything Module 7 has taught about benzodiazepine tapering carries over. The relationship between dose and receptor occupancy is hyperbolic, so reductions made by flat linear amounts cause increasingly large drops in effect near the end, which can mean increasingly severe withdrawal (p.485). Hyperbolic dose reductions are the method here too.

Why short half-lives change the picture

These drugs clear fast. Zaleplon has an elimination half-life of about 1 hour, so when taken once at night, plasma levels fall to zero before the next dose (p.480). Zolpidem has a mean half-life of 2.4 hours (range 0.7–3.5), and zopiclone about 5 hours (p.485, 490). Two consequences follow. First, dosing less often, including every-other-day dosing, is not recommended for tapering, because the swings in plasma level can provoke withdrawal (p.480, 485, 490). Second, inter-dose withdrawal can appear even mid-treatment. The zolpidem label notes withdrawal syndromes can occur with night-time doses (p.485). A patient who wakes at 3 a.m. anxious and wired may be experiencing the drug wearing off, not a new problem.

The as-needed (PRN) trap

Many patients use a Z-drug “only when I need it.” This sounds low-risk, and the intermittent pattern can hide dependence in plain sight. Here is the loop: the drug is taken, sleep follows, the body adapts a little, and on a night without it, sleep is worse than baseline. That rebound insomnia, which the Maudsley flags as a known withdrawal effect (most commonly insomnia with Z-drugs), reads to the patient as proof they still need the pill (p.481). So the “as-needed” use is partly being driven by withdrawal from the last dose. Naming this loop out loud, without blame, is often the first useful step. The fix is not more willpower on off-nights. It is a planned, gradual reduction that lets the sleep system re-regulate.

Building the taper

The Maudsley gives faster, moderate, and slower example schedules for each Z-drug, and is explicit that “none of these regimens should be seen as prescriptive” (p.482). Each reduction is made once the withdrawal symptoms from the previous step have largely resolved, so the drops don’t stack up. The book aims for a rate that produces tolerable symptoms abating within a week or two, with reductions about every 1–2 weeks for the Z-drugs (p.481, 487). The end of the taper matters most. Because of the hyperbolic curve, small final doses still carry meaningful receptor effect, so the last step before zero must be small. Liquids, suspensions made from a tablet or opened capsule, compounded smaller doses, or tapering strips make those tiny doses possible (p.481, 486–487, 491).

Switch to diazepam, or taper directly?

For benzodiazepines, switching to long-acting diazepam first is a common move. With Z-drugs the Maudsley leans the other way. Because these drugs have such short half-lives and are dosed once a day, “it might be that less physical dependence will be produced in response to this drug than diazepam,” so a direct taper is often easier than switching (p.483, 489, 493). The switch is reserved mainly for people reporting inter-dose withdrawal, for whom a longer-acting agent can smooth the day. The book gives equivalences for that route: 10 mg zaleplon ≈ 5 mg diazepam, 10 mg zolpidem ≈ 5 mg diazepam, and 7.5 mg zopiclone ≈ 5 mg diazepam (p.484, 489, 493).

Complex cases: combinations and long-term users

The hardest situations are combinations, a benzodiazepine plus a Z-drug, or a Z-drug stacked with other sedatives, and very long-term users. Two principles keep these safe. One change at a time. Reduce a single agent while holding the rest steady, so that if symptoms appear you know what caused them. Slow the pace, don’t rush the finish. Withdrawal tends to be more severe for people who have used these drugs for longer, at higher doses, and for those who have had withdrawal trouble before, all of which point toward smaller initial reductions and a longer overall plan (p.481). When even the slowest published schedule is too much, micro-tapering, very small daily reductions calculated by spreading a step’s change across 7 to 14 days, is the next gear down (p.488–489). A note on protracted cases: the existence of protracted withdrawal with Z-drugs has not been formally investigated, though it has been suggested there should be little difference from benzodiazepines given their similar pharmacology (p.485, 490). Believe the patient in front of you, hold the dose, and slow down.

Source: Horowitz MA, Taylor D. The Maudsley Deprescribing Guidelines (Wiley Blackwell, 2024), Ch.3 — zaleplon (p.480–484), zolpidem (p.485–489), and zopiclone (p.490–494) monographs.