gabapentin (Neurontin)

no boxed warning, and what the 2019 breathing warning says

the gabapentin label has no boxed warning. that is worth being accurate about, because the december 2019 fda action is often described as a black box and it was not; the fda required new wording in the warnings and precautions section, and required manufacturers to run clinical trials on abuse potential, particularly in combination with opioids [fda-2019]. the fda’s communication, dated december 19, 2019 and still live on fda.gov, says serious breathing difficulties may occur in people using gabapentin or pregabalin who have respiratory risk factors: opioids or other central nervous system depressants taken at the same time, conditions that reduce lung function such as copd, and older age [fda-2019]. it lists the symptoms to watch for: confusion, unusual dizziness, extreme sleepiness, slowed or shallow breathing, unresponsiveness, and bluish skin [fda-2019].

the label’s wording, section 5.8: "there is evidence from case reports, human studies, and animal studies associating gabapentin with serious, life-threatening, or fatal respiratory depression when coadministered with cns depressants, including opioids, or in the setting of underlying respiratory impairment." when the two are prescribed together, or gabapentin is prescribed to someone with impaired breathing, it says to monitor for slowed breathing and sedation and to consider starting at a low dose [label 5.8]. the interactions section adds that respiratory depression and sedation, "sometimes resulting in death", have been reported with morphine, hydrocodone, oxycodone and buprenorphine; morphine raises gabapentin blood levels, and gabapentin lowers hydrocodone exposure [label 7.1]. the overdose section records fatal respiratory depression with gabapentin alone and in combination with other cns depressants [label 10].

the size of the opioid risk comes from one study. among opioid users in ontario, 1,256 who died of an opioid-related cause were matched with 4,619 opioid-using controls; 12.3% of the deaths and 6.8% of the controls had been prescribed gabapentin in the prior 120 days, an adjusted odds ratio of 1.49 (95% ci 1.18 to 1.88). at moderate and high gabapentin doses (900 mg a day and above) the odds were about 60% higher. the same design found no association for a non-sedating painkiller, nsaids, which is what you would expect if the gabapentin finding is about sedation rather than about the kind of patient who takes a second drug [gomes-2017]. it is a case-control study in administrative data: association, not proof of cause, and ontario only [gomes-2017].

the suicidality warning that every antiepileptic carries

section 5.5 says antiepileptic drugs, "including gabapentin, increase the risk of suicidal thoughts or behavior in patients taking these drugs for any indication". the evidence is a pooled analysis of 199 placebo-controlled trials of 11 antiepileptic drugs: people randomised to a drug had about twice the risk of suicidal thinking or behaviour (adjusted relative risk 1.8, 95% ci 1.2 to 2.7). in absolute terms, 0.43% of 27,863 people on a drug against 0.24% of 16,029 on placebo, about one additional case for every 530 people treated over a median of 12 weeks. there were four suicides on drug and none on placebo, a number the label says is too small to allow any conclusion. the risk appeared as early as one week in and persisted through the trials, which mostly ended by 24 weeks [label 5.5].

the label breaks it down by why people were taking the drug, per 1,000 patients: epilepsy 3.4 on drug against 1.0 on placebo; psychiatric conditions 8.5 against 5.7; other conditions 1.8 against 1.0; in total 4.3 against 2.4, a difference of about two extra people per thousand [label 5.5, table 2]. two things to read carefully. this is a class-wide analysis, not a gabapentin-specific one; the label says the risk "was generally consistent among drugs" and applies it to all of them. and the label pairs it with the illness: anyone prescribing must weigh the risk "with the risk of untreated illness", since epilepsy and the psychiatric conditions these drugs are used for raise the risk of suicidal thoughts themselves [label 5.5]. suicidal behaviour and ideation have also been reported after gabapentin is stopped [label 5.5, 5.6].

dress, anaphylaxis and the other warnings, in plain words

the warnings section runs from 5.1 to 5.10. the april 2025 revision rewrote 5.5 and 5.6 and removed a former 5.10 on sudden unexplained death in epilepsy [label, recent major changes]. the rest:

• dress, or multiorgan hypersensitivity [5.1]: fever, rash or swollen lymph nodes together with organ involvement such as hepatitis, nephritis, blood abnormalities, myocarditis or myositis, "sometimes resembling an acute viral infection". the label says some of these reactions have been fatal or life-threatening, that early signs can appear before any rash, and that the drug should be stopped if no other cause is found.
• anaphylaxis and angioedema [5.2]: difficulty breathing and swelling of the lips, throat and tongue, "after the first dose or at any time during treatment"; stop the drug and seek immediate care.
• driving [5.3]: driving studies with a prodrug of gabapentin showed significant impairment; the label says people cannot reliably judge their own competence or how sleepy they are, and that how long the impairment lasts after starting is unknown.
• sleepiness, dizziness and unsteadiness [5.4]: in the epilepsy trials at up to 1,800 mg, somnolence 19% against 9% on placebo, dizziness 17% against 7%, ataxia 13% against 6%; in the shingles-pain trials at up to 3,600 mg, somnolence 21% against 5% and dizziness 28% against 8%. the label says to watch for cns depression when other sedating drugs are on board "because of potential synergy".
• stopping [5.6]: antiepileptic drugs "should not be abruptly discontinued"; when gabapentin is stopped, "the dose should be tapered over at least a one-week period". withdrawal symptoms have followed both short-term and long-term use, usually within 48 hours, and have included seizures, depression, suicidal ideation, agitation, confusion, psychotic symptoms, anxiety, insomnia, nausea, pain, sweating and tremor [5.6, 9.3].
• status epilepticus [5.7]: 0.6% on gabapentin against 0.5% on placebo in the controlled epilepsy trials; the label cannot say whether the drug changes the rate.
• children aged 3 to 12 [5.9]: emotional lability 6% against 1.3%, hostility 5.2% against 1.3%, hyperkinesia 4.7% against 2.9%, thought disorder 1.7% against 0.
• tumours [5.10]: gabapentin increased pancreatic tumours in rats; the label says the clinical significance is unknown and that the premarketing trials cannot tell whether the tumours seen in people were affected by treatment.
• kidneys and age [2.3, 8.5]: gabapentin is cleared almost entirely by the kidneys, so the dose is adjusted to kidney function, and older people, in whom kidney function falls, got a larger effect from the same dose in the pain trials; peripheral oedema and ataxia rose with age.
• antacids [7.3]: an aluminium and magnesium hydroxide antacid cut gabapentin absorption by about 20%; the label says to take gabapentin at least two hours after it.
• after marketing [6.2]: jaundice and raised liver tests, low sodium, rhabdomyolysis, movement disorders, breast enlargement, sexual side effects, and the skin reactions stevens-johnson syndrome and erythema multiforme have been reported, at rates the label cannot estimate.

misuse, dependence and withdrawal: what the label and the reviews say

gabapentin "is not a controlled substance" under federal law [label 9.1]. the label says misuse and abuse "have been reported in the postmarketing setting and published literature", that most of the people described had a history of polysubstance abuse, that some were taking more than the recommended dose for unapproved uses, and that "the abuse potential of gabapentin has not been evaluated in human studies" [label 9.2]. it asks prescribers to check for a history of drug abuse and to watch for self-escalation of the dose and drug-seeking [label 9.2]. the dependence section says the same of dependence potential: not evaluated in human studies, but withdrawal symptoms observed after short- and long-term treatment, usually within 48 hours of stopping [label 9.3].

the published literature the label refers to has been reviewed three times by the same group. the 2021 update found 55 new studies in four years and concluded that gabapentinoids "are increasingly abused or misused to self-medicate", that they "can produce desirable effects alone but are often used concomitantly with other drugs", that opioid use disorder "is the greatest risk factor" for their abuse, and, "the most concerning finding", growing evidence of patient harm including hospital use and opioid-overdose mortality [evoy-2021]. the 2026 update, 100 further studies to january 2025, found "more compelling evidence that gabapentinoids increase overdose risk, particularly when used with opioids or benzodiazepines", with misuse newly documented in adolescents and in people with chronic pain [evoy-2026]. the uk guideline on medicines associated with dependence lists gabapentinoids alongside opioids, benzodiazepines, z-drugs and antidepressants, says withdrawal "may take several months or more", and asks for a written taper; gabapentinoids are the one class it reduces by a fixed amount at each step rather than proportionally [nice-2022].

what the approval trials measured, and for how long

the shingles-pain indication rests on two randomised placebo-controlled trials in 563 people with pain lasting more than three months after the rash healed: one 8-week trial at 3,600 mg a day (113 on gabapentin, 116 on placebo) and one 7-week trial at 1,800 and 2,400 mg (223 against 111). pain was scored daily on a 0 to 10 scale; both trials beat placebo at every dose, with the difference visible by week one, and the label says the extra benefit of going above 1,800 mg "was not demonstrated" [label 14.1, 2.1].

the seizure indication rests on three add-on trials in 705 people aged 12 and over and one in 247 children aged 3 to 12, each with a 12-week treatment period after a 12-week baseline, in people still having at least two to four seizures a month on their existing drugs. the outcome was the share of people whose seizures fell by half: 23% against 9% at 1,200 mg in the first trial; 16% against 8% at 1,200 mg in the second, which was not statistically significant, with 26% at 1,800 mg, which was; 22% against 10% at 900 mg in the third. in the children, the response ratio was significant but the responder rate (21% against 18%) was not, and a trial in infants aged one month to three years found no difference [label 14.2]. nothing on the label is placebo-controlled beyond 12 weeks, and nothing on it tested a psychiatric condition.

side effects, with the label’s numbers

in the shingles-pain trials (336 on gabapentin at up to 3,600 mg against 227 on placebo): dizziness 28% against 8%, somnolence 21% against 5%, peripheral oedema 8% against 2%, weakness 6% against 5%, diarrhea 6% against 3%, dry mouth 5% against 1%, constipation 4% against 2%, ataxia 3% against 0, abnormal thinking 3% against 0, blurred vision 3% against 1%, weight gain 2% against 0. 16% of people on gabapentin and 9% on placebo stopped because of a side effect, most often dizziness, somnolence and nausea [label 6.1, table 3].

in the add-on epilepsy trials in people over 12 (543 against 378): somnolence 19% against 9%, dizziness 17% against 7%, ataxia 13% against 6%, fatigue 11% against 5%, nystagmus 8% against 4%, tremor 7% against 3%, double vision 6% against 2%, blurred vision 4% against 1%, weight gain 3% against 2%, amnesia 2% against 0, depression 2% against 1%, impotence 2% against 1%. somnolence and ataxia rose with dose. about 7% of people in the premarketing trials stopped for a side effect [label 6.1, table 4]. in children aged 3 to 12 (119 against 128): viral infection 11% against 3%, fever 10% against 3%, nausea or vomiting 8% against 7%, somnolence 8% against 5%, hostility 8% against 2% [label 6.1, table 5].

anxiety and sleep: not an approved use, and what the trials say

gabapentin for anxiety, insomnia or mood is not an approved use. the fda has never reviewed evidence for it, the label contains no trial of it, and the suicidality warning above applies to the drug "for any indication" [label 1, 5.5]. the evidence that exists is published trials, and it points both ways.

in favour. the largest synthesis, a 2022 systematic review of 55 double-blind randomised trials and 15 open-label studies of gabapentin and pregabalin in bipolar disorder, anxiety and insomnia, found "a consistent but not universal effect in favour of gabapentinoids compared to placebo" across the anxiety spectrum (standardised mean differences from −2.25 to −0.25), with the clearest signal in anxiety before surgery (gabapentin −0.92, 95% ci −1.32 to −0.52), and concluded there is "moderate evidence of the efficacy of gabapentinoids in anxiety states" [hong-2022]. the best-known single trial randomised 69 people with social phobia to gabapentin at 900 to 3,600 mg a day or placebo for 14 weeks and found a significant reduction in symptoms on clinician and patient scales, with more dizziness, dry mouth, somnolence, nausea and decreased libido on the drug [pande-1999]. a 2015 systematic review found that of 34 trials of adequate quality, those in some anxiety disorders and in alcohol craving and withdrawal showed benefit [berlin-2015].

against, or smaller than it sounds. the same 2022 review found "minimal evidence in bipolar disorder and insomnia" (four double-blind trials in bipolar disorder, none poolable; insomnia "inconclusive") and said the drugs "should be used for these disorders only with strong justification" [hong-2022]. the panic-disorder trial from the same programme, 103 people over 8 weeks, found no difference from placebo on the primary scale (p = 0.606); the reported benefit in the more severely ill is a post hoc analysis [pande-2000]. the 2015 review noted "there are no studies for generalized anxiety disorder", the condition it is most often prescribed for, and no clear evidence in depression, ptsd prevention or ocd [berlin-2015]; a 2019 review of 54 papers reached the same shape, benefit in preoperative anxiety and social phobia, little as a sole treatment, and warned that many of the positive studies were open-label [ahmed-2019]. a 2019 clinical overview of off-label gabapentinoid prescribing concluded that "review articles and guidelines tend to overstate gabapentinoid effectiveness" and that prescribers "should acknowledge to patients that potential benefits are uncertain for most off-label uses" [goodman-2019].

who paid. the 2022 review was funded by the uk department of health; two of its authors hold an unrestricted educational grant from johnson & johnson on unrelated work, and one has received fees from angelini pharma [hong-2022]. the 1999 and 2000 trials predate routine funding statements and the abstracts read for this page carry none, so their funding is recorded as unknown rather than inferred [pande-1999; pande-2000]. the 2015 review was supported by a national institute on aging grant [berlin-2015].

what the research says, in both directions

in favour. the approval trials show a real effect on refractory partial seizures and on post-shingles pain, visible in the first week [label 14.1, 14.2]. in anxiety, the pooled randomised evidence is moderate and mostly positive, strongest before surgery and in social phobia [hong-2022; pande-1999]. the respiratory warning is specific: the label and the fda tie it to co-prescribed depressants and to existing lung disease, not to healthy people taking gabapentin alone [label 5.8; fda-2019].

against. the drug has no approved psychiatric use and its longest controlled evidence is 12 weeks [label 14]. the panic trial was negative, generalised anxiety has no trial, insomnia is inconclusive, and the people who reviewed the off-label literature most carefully say its effectiveness is overstated [pande-2000; berlin-2015; hong-2022; goodman-2019]. the misuse literature has grown from 59 studies to more than 200 in nine years, and the harm it documents concentrates where gabapentin is stacked on opioids or benzodiazepines [evoy-2021; evoy-2026].

the warnings, measured. the breathing warning’s size is the ontario study: about one and a half times the odds of an opioid-related death with a gabapentin prescription in the prior four months, rising with dose [gomes-2017]. the suicidality warning’s size is the label’s own pooled analysis: about two extra people per thousand with suicidal thoughts or behaviour, over about 12 weeks, across the whole drug class [label 5.5]. the dependence warning’s timescale is in the uk guideline [nice-2022]. gabapentin has a legal and safety record page on this site, which keeps the court and regulatory record separately from the label facts here [legal].

questions people ask

does gabapentin (neurontin) have a black box warning?

no. the gabapentin label has no boxed warning. in december 2019 the fda added a warning to the warnings and precautions section, not a box, saying that serious and potentially fatal breathing problems can occur in people taking gabapentin or pregabalin who have risk factors: opioids or other central nervous system depressants taken at the same time, lung conditions such as copd, and older age. the fda also required manufacturers to run trials on abuse potential, particularly with opioids. the label carries nine other numbered warnings, including the class-wide suicidality warning for antiepileptic drugs and a multiorgan hypersensitivity reaction (dress) that has been fatal.

can you take gabapentin with opioids?

the label does not forbid it, but it says that evidence from case reports, human and animal studies links gabapentin with serious, life-threatening or fatal respiratory depression when taken with cns depressants including opioids, and that respiratory depression and sedation, sometimes resulting in death, have been reported with morphine, hydrocodone, oxycodone and buprenorphine. when the two are prescribed together it asks for monitoring for slowed breathing and sedation and a low starting dose. in a study of opioid users in ontario, those who died of an opioid-related cause were about one and a half times as likely to have had a gabapentin prescription in the prior four months, rising with dose. the combination is a question for whoever prescribes either drug, and both need to know about the other.

is gabapentin a controlled substance, and is it addictive?

gabapentin is not a federally controlled substance; the label says so in section 9.1. the same label says misuse and abuse have been reported after marketing and in the published literature, mostly in people with a history of using several substances, sometimes at doses above the recommended ones for unapproved uses, and that its abuse potential "has not been evaluated in human studies". physical dependence is a separate thing: the label says withdrawal symptoms have followed short-term and long-term use, usually within 48 hours of stopping. systematic reviews of the misuse literature, now more than 200 studies, find that opioid use disorder is the biggest risk factor for gabapentinoid misuse and that overdose risk rises when they are combined with opioids or benzodiazepines.

does gabapentin work for anxiety?

anxiety is not an approved use; the fda has never reviewed it and the label contains no trial of it. the published trials point both ways. a 2022 systematic review of 55 double-blind trials found moderate evidence that gabapentinoids reduce anxiety against placebo, clearest for anxiety before surgery and in social phobia, where a 69-person 14-week trial was positive. the panic-disorder trial, 103 people over 8 weeks, found no difference from placebo. there are no trials in generalised anxiety disorder, insomnia evidence is inconclusive, and the reviewers say gabapentinoids should be used for these conditions "only with strong justification". whatever the reason for a prescription, the label’s suicidality warning applies to any indication.

what happens if you stop gabapentin suddenly?

the label says antiepileptic drugs should not be stopped abruptly because seizures can become more frequent, and that gabapentin should be tapered over at least one week. withdrawal symptoms have been observed after both short and long courses, usually within 48 hours, and have included seizures, depression, suicidal ideation and behaviour, agitation, confusion, disorientation, psychotic symptoms, anxiety, insomnia, nausea, pain, sweating, tremor, headache and dizziness. the uk guideline on dependence-forming medicines lists gabapentinoids with opioids and benzodiazepines and says withdrawal may take several months or more. a taper plan belongs with a prescriber.

what are the most common gabapentin side effects?

in the label’s post-shingles pain trials at up to 3,600 mg a day: dizziness 28% vs 8% on placebo, sleepiness 21% vs 5%, swollen ankles 8% vs 2%, diarrhea 6% vs 3%, dry mouth 5% vs 1%. in the add-on epilepsy trials: sleepiness 19% vs 9%, dizziness 17% vs 7%, unsteadiness 13% vs 6%, fatigue 11% vs 5%, tremor 7% vs 3%, double vision 6% vs 2%. sleepiness and unsteadiness rose with dose, and the label says not to drive until you know how it affects you, since driving studies showed significant impairment and people misjudge their own sleepiness. in children aged 3 to 12 the label adds hostility (8% vs 2%) and emotional lability.

sources for this section

the library rows link to our one-page summary of each study, with the original paper, its funding and our evidence grade on that page.

1. gabapentin capsules prescribing information, fda label on dailymed (set id 4615aaa7-73d4-4473-8f26-e7a982f8a166, version 1, effective august 27, 2026): sections 1, 2.1 to 2.3, 5.1 to 5.10, 6.1 tables 3 to 5, 6.2, 7.1, 7.3, 8.5, 9.1 to 9.3, 10, 14.1, 14.2, recent major changes
2. drugs@fda, nda 020235 (neurontin), approved december 30, 1993; the approval record the dossier above cites
3. ahrq medical expenditure panel survey, prescribed medicines public-use files 2014 (hc-168a) and 2024 (hc-254a): people with at least one gabapentin purchase, weighted; computed by veisund, method on the usage section of this page
4. fda drug safety communication, december 19, 2019: fda warns about serious breathing problems with seizure and nerve pain medicines gabapentin (neurontin, gralise, horizant) and pregabalin (lyrica, lyrica cr); page re-read live on fda.gov on the verification date (library row)
5. the neurontin (brand) label’s sections 5.6 and 5.8 quoted, dailymed set id 97935fd9-1d4a-43b6-a5d9-de994591187b, spl version 48, august 2026; the same wording as the generic label read for this section (library row)
6. gomes t, juurlink dn, antoniou t, mamdani mm, paterson jm, van den brink w. gabapentin, opioids, and the risk of opioid-related death: a population-based nested case-control study. plos medicine 2017 (library row)
7. national institute for health and care excellence. medicines associated with dependence or withdrawal symptoms: safe prescribing and withdrawal management for adults (ng215), 2022 (library row)
8. hong jsw, atkinson lz, al-juffali n, awad a, geddes jr, tunbridge em, et al. gabapentin and pregabalin in bipolar disorder, anxiety states, and insomnia: systematic review, meta-analysis, and rationale. molecular psychiatry 2022;27:1339-1349 (pubmed 34819636; abstract and declarations read on the verification date; not yet a library row)
9. pande ac, davidson jr, jefferson jw, janney ca, katzelnick dj, weisler rh, et al. treatment of social phobia with gabapentin: a placebo-controlled study. journal of clinical psychopharmacology 1999;19:341-348 (pubmed 10440462; abstract read; funding statement not in the record)
10. pande ac, pollack mh, crockatt j, greiner m, chouinard g, lydiard rb, et al. placebo-controlled study of gabapentin treatment of panic disorder. journal of clinical psychopharmacology 2000;20:467-471 (pubmed 10917408; abstract read; funding statement not in the record)
11. berlin rk, butler pm, perloff md. gabapentin therapy in psychiatric disorders: a systematic review. primary care companion for cns disorders 2015;17(5) (pubmed 26835178; abstract read; national institute on aging grant on the record)
12. ahmed s, bachu r, kotapati p, adnan m, ahmed r, farooq u, et al. use of gabapentin in the treatment of substance use and psychiatric disorders: a systematic review. frontiers in psychiatry 2019;10:228 (pubmed 31133886; abstract read)
13. goodman cw, brett as. a clinical overview of off-label use of gabapentinoid drugs. jama internal medicine 2019;179:695-701 (pubmed 30907944; abstract read)
14. evoy ke, sadrameli s, contreras j, covvey jr, peckham am, morrison mn. abuse and misuse of pregabalin and gabapentin: a systematic review update. drugs 2021;81:125-156 (pubmed 33215352; abstract read)
15. evoy ke, drain d, pedigo j, covvey jr, ramey o. misuse of pregabalin and gabapentin: a second systematic review update. drugs 2026;86:1365-1378 (pubmed 42236659; abstract and declarations read; authors declare no relevant conflicts)
16. gabapentin: legal and safety record on this site (court and regulatory record, kept separately from the label facts above)

Most-reported reactions

• Drug ineffective29,638
• Fatigue24,972
• Nausea22,728
• Pain21,181
• Off label use19,807
• Diarrhoea18,114
• Headache17,897
• Dizziness16,238
• Fall15,536
• Dyspnoea14,256