How do you taper long-term benzodiazepines?

We return to Mr. Davis, a 75-yo man who has taken temazepam 30mg at bedtime for 15 years for insomnia. He recently fell and fractured his wrist. You discussed the risks and benefits of continuing vs stopping temazepam and the rationale for tapering the medication as described in the prior PsychSnap. Mr. Davis decides to move forward with a taper of temazepam.

How do you taper long-term benzodiazepines?

What follows is a set of strategies rather than a fixed, universal protocol.  They are designed to prevent the experience of physical withdrawal or symptom rebound (i.e. anxiety or insomnia) when stopping long-term benzodiazepines. The strategies for tapering benzodiazepines in someone with a benzodiazepine use disorder or an urgent medical need to stop benzodiazepines, like delirium, may be different.*

Tapering Principles 

  • Go slow – Many published tapering schedules for benzodiazepines last 6 weeks. While this is medically safe (i.e. people are unlikely to have seizures), many people will need longer tapering schedules. I generally taper over 3-6 months, adjusting the speed as necessary based on the patient’s symptoms and life circumstances.
  • Stop slow as you go low – The hardest part of the taper will always be the last few weeks to months. Patients can tolerate larger dose reductions at the beginning of a taper than at the end. The figure below from Horowitz et al. helps us understand why. 

Benzodiazepines are positive allosteric modulators at the GABA-A receptor. As the benzodiazepine (diazepam) dose increases, the GABA occupancy increases, but it does NOT increase linearly. In fact, by the time we reach a dose of diazepam 4mg (approximately temazepam 8mg or lorazepam 0.5mg), more than 80% of the GABA-A receptor is occupied. Put another way, the drop in benzodiazepine-GABA-A receptor binding from diazepam 4mg to 0.5mg (3.5mg) is similar to the drop from diazepam 0.5mg to 0mg (0.5mg). One easy way to “stop slow as you go low” is to reduce the dose by a certain percentage of the current dosage. As the dose gets lower, the size of the decrease gets smaller. Expert opinion suggests a dose reduction of 10-25% every 1-2 weeks at the beginning, and 10% over a month at the end.

  • Keep going – People generally decrease benzodiazepine doses every 1-2 weeks during a taper. Scheduling visits at a similar frequency, particularly at the start of a taper, can help us understand how a patient is feeling and figure out if now is the time to decrease the dose further. If people are experiencing withdrawal symptoms (irritability, insomnia, anxiety, tremor, sweating), I wait for them to feel ok for at least a week before decreasing the dose further. If a patient feels that a taper is too fast, we can pause it. We make an agreement ahead of time that we won’t increase the dose again once we’ve gone down, and there will be no early refills if patients take more pills than prescribed. Shorter prescriptions can be helpful if people are running out early. 
  • Work around pill sizes – What if we want to decrease a dose by 10% but the available pills don’t come in that dose? For example, the smallest available capsule of temazepam is 7.5mg. The smallest decrease in dose that we can accomplish with available pills for Mr. Davis is from temazepam 30mg to temazepam 22.5mg, a 25% decrease in dose. This issue comes up when tapering antidepressant and antipsychotic medications as well. Your tools include cutting tablets, liquid formulations (available for many benzodiazepines, including lorazepam, clonazepam, and diazepam), compounded pills from a compounding pharmacy, or switching to another benzodiazepine.
  • To switch or not to switch – Some people recommend consolidating all current benzodiazepines to diazepam equivalents, sometimes over a number of weeks to make the transition less disruptive because the conversion between different benzodiazepines is inexact. Then diazepam can then be tapered over time. In expert opinion, diazepam is commonly preferred because it is long-acting and very slowly metabolized, allowing for a smooth gradual fall in benzodiazepine blood levels over time. Diazepam is also available in 2mg tabs that can be broken in half and a liquid formulation that allows for very small dose changes at the end. The conversion to diazepam is particularly useful if the current benzodiazepine is short acting, like alprazolam. The downsides of this approach are that long-acting diazepam metabolites can build up, particularly in the elderly, and it may not be necessary for some patients. For a patient on a stable daily dose of a single benzodiazepine that is not short acting, I offer the patient the choice to switch to diazepam vs tapering with the current benzodiazepine. If they don’t have a preference, I generally taper with the current benzodiazepine and only switch to diazepam if we get stuck.
  • Adjuvant medications – Adjuvant medications may be helpful during benzodiazepine withdrawal, but no medications have consistently demonstrated relief of general benzodiazepine withdrawal. Some experts suggest using anti-epileptic drugs like carbamazepine or oxcarbazepine during the course of the taper and for a month or two after to prevent seizures, particularly in people with benzodiazepine use disorders who have been taking variable doses of benzodiazepines or may be undergoing more rapid tapers.
  • Increase psychological support – Ideally, a benzodiazepine taper occurs during a time of low psychosocial stress with some increased support. If a benzodiazepine was started to treat acute anxiety or insomnia during a stressful time that has since passed, the support of frequent visits with you during the taper may suffice. A patient with a chronic anxiety or insomnia disorder may benefit from concurrent psychotherapy for the underlying problem.
  • Write it down – Many people don’t take pills as prescribed and keeping track of changing doses adds an extra layer of complexity. Frequent visits can be helpful to simplify instructions. This can also be a good time to confirm patient literacy. I like the question, “How comfortable are you with how you read?” If they feel comfortable with reading, I write out the taper week by week. If they don’t, I have them bring in the pill bottles, and we tape the pills to a piece of paper to create a visual representation.

Returning to Mr. Davis, our 75 year old man who is taking temazepam 30mg and recently fell. Temazepam is a short-to-intermediate acting benzodiazepine, with a half- life elimination of 4-18 hours, probably on the longer side in someone who is older. It is available in capsules only, with 7.5mg, 15mg, 22.5mg, and 30mg dosages available. Given the limited pill dosages available, the smallest initial dose reduction with pills would be from 30mg to 22.5mg nightly – a 25% reduction. Mr. Davis is open to trying this larger decrease and tolerates the change without problems. At your next visit 3 weeks later, you discuss the potential next steps. The smallest decrease in dosage that can be done with available pills at this point is a 33% reduction from 22.5mg to 15mg. Given how well he tolerated the previous decrease, Mr. Davis wants to try reducing the temazepam dose from 22.5mg to 15mg. He calls 5 days later to say that since reducing the temazepam dose, it has been taking him hours to fall asleep, despite using all of the strategies that he has learned in CBT-i. In other words, we’ve gotten stuck and need to course correct.  

We talk again about the risks and benefits of switching to diazepam vs using compounded temazepam pills that allow for smaller dose decreases. He calls a compounding pharmacy and learns that they do not accept insurance. The compounded pills will cost him around $100/month. He decides to switch to diazepam. Temazepam 22.5mg nightly is the last dose that he tolerated. Benzodiazepine equivalence tables help us estimate roughly equivalent doses between different benzodiazepines. The roughly equivalent dose of temazepam 22.5mg is diazepam 11.25mg, so we plan to switch to diazepam 11mg bedtime. 

Diazepam is a long acting benzodiazepine, with a half-life elimination of 20-100 hours. It is available in 10mg, 5mg, and 2mg pills, in addition to liquid formulations that allow for much smaller doses (0.75mg, 0.5mg, etc). When we switch directly from temazepam 22.5mg to diazepam 11mg,** we know that this is only an estimate of equivalency, so we monitor Mr. Davis for oversedation and the symptoms of benzodiazepine withdrawal that would indicate that we over- or under- estimated the diazepam dose. Mr. Davis has no trouble with the conversion to diazepam and has been sleeping well. After 2 weeks on diazepam 11mg, he decreases the diazepam by 1mg every 2 weeks as instructed and successfully stops diazepam. The entire process took 4 months. 

*There is data emerging for the use of gabapentin, phenobarbital, or valproate with clonidine instead of a benzodiazepine taper. These approaches are not the standard of care at this time.

**A cross-taper with overlapping doses of a shorter-acting benzodiazepine and a long-acting benzodiazepine over a number of weeks can be used instead of a direct switch. The cross-taper is more complicated than a direct switch, but it may make the transition between the two different medications smoother and allow for course corrections within the process.  This can be particularly helpful for people on high dose benzodiazepines.

Key Points

  1. Strategies for tapering chronic benzodiazepines include: go slow, stop slow as you go low, consider switching to diazepam, provide additional psychological support, and write it down. 
  2. People can generally tolerate a larger dose reduction at the beginning of a benzodiazepine taper than at the end.
  3. It is appropriate to pause a benzodiazepine taper or course correct if withdrawal symptoms are present or something doesn’t go according to plan.


Ashton, Heather. The diagnosis and management of benzodiazepine dependence. Current Opinion in Psychiatry 18(3):p 249-255, May 2005.

Horowitz, Mark Abie, and David Taylor. “Tapering of SSRI treatment to mitigate withdrawal symptoms.” The Lancet Psychiatry 6.6 (2019): 538-546.

Send a Comment