
Valproate (valproic acid, divalproex)
Valproate is a first-choice drug for generalized epilepsy and works well for childhood absence seizures, acute mania and migraine prevention. Its serious risks are rare but real: liver failure (highest in young children and people with certain mitochondrial disorders), pancreatitis, high ammonia and, in pregnancy, a high rate of birth defects and lower IQ in children. Aspirin, CBD and changes in carnitine matter for supplement users.
Prescription medication: not a dietary supplement
This is an FDA-approved (or investigational) drug, not a supplement. It requires a prescription and medical supervision. The information below summarizes clinical-trial evidence for education only. It is not a recommendation to obtain or use it without a doctor.
Quick decision guide
May help most
People with generalized or absence epilepsy, focal seizures, acute bipolar mania or frequent migraines whose prescriber has judged valproate suitable, and who can keep up liver and blood tests
Common dosing range
Set by your prescriber, often guided by blood levels. Depakote label: mania starts at 750 mg a day in divided doses; epilepsy starts at 10 to 15 mg per kg a day; migraine starts at 250 mg twice daily, up to 1,000 mg a day. Maximum 60 mg per kg a day
When to expect effects
Mania trials lasted 3 weeks, with peak levels usually reached within 14 days; epilepsy doses are raised weekly; migraine prevention is judged over weeks to months
Watch out for
Boxed warnings for liver failure, harm to an unborn baby and pancreatitis. Not for pregnancy or people who could become pregnant unless other medicines have failed. Can raise ammonia and lower carnitine; ask your doctor whether to check them if you feel unusually tired, confused or sick to your stomach
What is it
Valproate is a prescription medicine for epilepsy, bipolar mania and migraine prevention. It is sold as valproic acid capsules and oral solution, and as divalproex sodium (Depakote delayed-release tablets, Depakote ER extended-release tablets and Depakote Sprinkles capsules) and generics; all deliver the same active valproate in the body. The Depakote label approves it for manic episodes of bipolar disorder, complex partial (focal) seizures and absence seizures, and to prevent migraine headaches. It carries boxed warnings for liver failure, harm to an unborn baby and pancreatitis.
Is it worth it for you?
Use this as a quick fit check, not a diagnosis.
Worth considering if…
Probably skip if…
Evidence at a glance
| Goal | Effect | Best fit | Time |
|---|---|---|---|
generalized and hard-to-classify epilepsy Good Evidence | Moderate; 12-month remission reached sooner than with lamotrigine (HR 0.76 in SANAD) | Children and adults with idiopathic generalized epilepsy | Doses are raised weekly; trials measured 12-month remission over several years |
childhood absence epilepsy Good Evidence | Large versus lamotrigine (58% vs 29% free of failure); similar to ethosuximide | Children with newly diagnosed childhood absence epilepsy | Assessed at 16 weeks |
acute manic episodes of bipolar disorder Good Evidence | Moderate; responders 48% vs 25% with placebo, similar to lithium | Adults hospitalized for acute mania, including those who did not respond to lithium | Within the 3-week trial; label says peak blood levels are usually reached within 14 days |
preventing episodic migraine in adults Good Evidence | Moderate; RR 2.18 for a 50% drop in headache frequency, NNT 4 | Adults with frequent episodic migraine who cannot become pregnant or use effective contraception | Judged over weeks to months of daily use |
generalized and hard-to-classify epilepsy
- Effect
- Moderate; 12-month remission reached sooner than with lamotrigine (HR 0.76 in SANAD)
- Best fit
- Children and adults with idiopathic generalized epilepsy
- Time
- Doses are raised weekly; trials measured 12-month remission over several years
childhood absence epilepsy
- Effect
- Large versus lamotrigine (58% vs 29% free of failure); similar to ethosuximide
- Best fit
- Children with newly diagnosed childhood absence epilepsy
- Time
- Assessed at 16 weeks
acute manic episodes of bipolar disorder
- Effect
- Moderate; responders 48% vs 25% with placebo, similar to lithium
- Best fit
- Adults hospitalized for acute mania, including those who did not respond to lithium
- Time
- Within the 3-week trial; label says peak blood levels are usually reached within 14 days
preventing episodic migraine in adults
- Effect
- Moderate; RR 2.18 for a 50% drop in headache frequency, NNT 4
- Best fit
- Adults with frequent episodic migraine who cannot become pregnant or use effective contraception
- Time
- Judged over weeks to months of daily use
Evidence for 4 uses
AI-assisted evidence assessment. Talk to your doctor before relying on any single supplement.
generalized and hard-to-classify epilepsy
Disease adjunctIn the SANAD trial, 716 UK children and adults with generalized or unclassifiable epilepsy were randomly assigned to valproate, lamotrigine or topiramate and followed for up to about 7 years. Valproate was significantly better than lamotrigine at reaching 12 months without seizures (HR 0.76) and was stopped less often than topiramate (HR 1.57). A 2022 Cochrane network meta-analysis of individual data found that no drug did better than valproate for generalized tonic-clonic seizures, and that valproate, lamotrigine and levetiracetam did not differ in treatment failure, with lamotrigine and levetiracetam the closest alternatives; it rated this evidence moderate certainty. SANAD was not blinded, and the authors noted that pregnancy risks must be weighed for women of childbearing age. The US label lists absence and complex partial seizures, not generalized tonic-clonic seizures by name.
Bottom line: Valproate is a first choice for generalized epilepsy; lamotrigine and levetiracetam are the closest alternatives.
Evidence is mixed
In SANAD, valproate reached 12-month remission sooner than lamotrigine (HR 0.76), but the 2022 Cochrane network meta-analysis found no difference between valproate, lamotrigine and levetiracetam in treatment failure for generalized seizures (moderate certainty).
childhood absence epilepsy
Disease adjunctIn a double-blind trial, 453 children with newly diagnosed childhood absence epilepsy were randomly assigned to ethosuximide, valproic acid or lamotrigine. After 16 weeks, 58% on valproic acid and 53% on ethosuximide were still free of treatment failure, compared with 29% on lamotrigine. Problems with attention were more common with valproic acid than ethosuximide (49% vs 33%), and the authors favored ethosuximide for that reason. This is one large trial with 16 weeks of follow-up for its main result.
Bottom line: Valproate controls childhood absence seizures as well as ethosuximide and far better than lamotrigine.
acute manic episodes of bipolar disorder
Disease adjunctIn a 3-week double-blind trial, 179 hospitalized adults with acute mania were randomly assigned to divalproex, lithium or placebo. At least 50% improvement in mania scores was seen in 48% on divalproex and 49% on lithium, compared with 25% on placebo, and divalproex worked about as well in people who had not responded to lithium before. The label notes that benefit beyond 3 weeks has not been shown in controlled trials. This is a single trial.
Bottom line: Divalproex roughly doubled the chance of improving acute mania compared with placebo.
preventing episodic migraine in adults
Disease adjunctA 2013 Cochrane review of 10 trials in adults with episodic migraine found that divalproex more than doubled the share of people whose headaches fell by at least half compared with placebo (4 trials, 542 participants; RR 2.18, about 1 extra responder for every 4 treated). Sodium valproate cut headaches by about 4 per 28 days versus placebo in 2 small trials (63 participants). It did not clearly beat flunarizine or propranolol in small head-to-head trials. Valproate does not treat a migraine attack once it starts, and the label forbids this use in pregnancy and in women who could become pregnant without effective contraception.
Bottom line: Valproate roughly doubles the chance of halving migraine frequency.
How it works
How to take it
What to track
Bottom line: Your prescriber sets the dose and the liver and blood tests; never stop or change it on your own.
Safety
Know the common side effects, key cautions, and who should avoid it.
Common side effects
Serious risks
Boxed warning: liver failure, sometimes fatal, usually in the first 6 months. Children under 2 and people with mitochondrial disorders caused by POLG gene mutations are at much higher risk. Liver tests are done before treatment and often during the first 6 months.
Boxed warning: fetal risk. Valproate can cause major birth defects, especially neural tube defects such as spina bifida, and lower IQ and neurodevelopmental disorders after exposure in the womb.
Boxed warning: life-threatening pancreatitis in children and adults, sometimes progressing rapidly; it can occur soon after starting or after years. Stomach pain, nausea, vomiting or loss of appetite need prompt medical review.
High blood ammonia and ammonia-related brain swelling (encephalopathy), which can occur with normal liver tests and is more likely with topiramate or an underlying urea cycle disorder; low body temperature can be a sign.
Low platelets and clotting problems; in one epilepsy trial, 27% of patients on about 50 mg per kg a day had a platelet count of 75 or below at least once.
Suicidal thoughts or behavior (a risk across anti-seizure drugs), DRESS and other serious allergic or skin reactions, and excessive sleepiness with dehydration and weight loss in older adults.
Who should avoid it
- People with liver disease or significant liver dysfunction
- People with mitochondrial disorders caused by POLG mutations (such as Alpers-Huttenlocher syndrome), and children under 2 suspected of having one
- People with a known urea cycle disorder, or an allergy to divalproex, sodium valproate or valproic acid
- For migraine prevention: anyone pregnant, and women who could become pregnant and are not using effective contraception
Pregnancy & breastfeeding
Pregnancy: valproate should not be used for epilepsy or bipolar disorder in women who are pregnant or planning pregnancy, or in women who could become pregnant, unless other medicines have failed or are unacceptable, and then with effective contraception; for migraine prevention it is contraindicated in pregnancy. The label reports a major malformation rate of 9 to 11% with about 1,000 mg a day, up to five times the rate with other anti-seizure monotherapies, a spina bifida risk of about 1 to 2%, and lower IQ at age 6 (97 versus 105 to 108 with other drugs) along with higher risks of autism and ADHD. Folic acid before conception and during pregnancy is routinely recommended, though it is not known whether it lowers these risks with valproate. Do not stop it suddenly if you become pregnant: plan any change with your prescriber. Breastfeeding: valproate passes into breast milk at 1 to 10% of the mother's blood level; the label asks that breastfed infants be watched for jaundice and unusual bruising or bleeding, and the decision is made with your prescriber.
Interactions
Valproate binds carnitine and can lower it; the label lists decreased carnitine among reported effects, and high blood ammonia can occur even with normal liver tests. In one study, plasma carnitine in a patient starting valproate dropped for 1 to 3 weeks and recovered by 3 to 5 weeks, and 10 long-term users had levels similar to controls, so low carnitine is not universal. In a case report, 1 g a day of carnitine lowered ammonia in a woman with valproate-related high ammonia; ask your prescriber whether to check your ammonia or carnitine, and use carnitine only under their supervision.
Aspirin pushes valproate off its binding proteins and slows its breakdown: in 6 children given fever-reducing doses, the free (active) fraction of valproate rose 4-fold. The standard total valproate blood test may not show this. Valproate can also lower platelets, and the label advises caution with the combination; talk to your prescriber or pharmacist before using aspirin, and do not stop aspirin a doctor prescribed without asking them.
In trials of prescription cannabidiol (Epidiolex), most liver enzyme rises happened in people also taking valproate: in Lennox-Gastaut and Dravet syndrome trials (10 to 20 mg per kg a day), ALT above 3 times normal occurred in 21% taking valproate without clobazam versus 3% taking neither drug. After approval, high ammonia was also reported, mostly in people taking valproate, clobazam or both. These data come from prescription doses; talk to your neurologist before using any CBD product, as liver tests may be needed.
The label warns that topiramate with valproate has caused high blood ammonia, with or without brain effects, and low body temperature, even in people who tolerated each drug alone. Report unusual drowsiness, vomiting, confusion or feeling cold to your prescriber.
These hospital antibiotics can drop valproate below effective levels and lead to loss of seizure control. Make sure the treating team knows you take valproate so levels can be checked or another antibiotic chosen.
The Depakote Medication Guide says not to drink alcohol while taking it, because the two can affect each other and cause side effects such as sleepiness and dizziness, and the label warns valproate's CNS depression is more likely with another CNS depressant such as alcohol. The label also asks you to tell your provider if you drink before starting, though it does not state that alcohol raises valproate's liver risk. Avoid alcohol, do not drive until you know how valproate affects you, and tell your prescriber if avoiding alcohol is hard.
In 75 children on valproate alone, activity of biotinidase (the enzyme that recycles biotin) was much lower at moderate and high doses than in healthy children, and rash and hair loss improved after 10 mg a day of biotin in this uncontrolled study. Hair loss is a common valproate side effect, but evidence that biotin helps is limited; ask your prescriber before trying it. If you take high-dose biotin, tell the lab before blood tests, because 10 mg a day distorted some hormone tests in healthy adults.
Estrogen-containing contraceptives can speed valproate clearance, lowering its level and possibly increasing seizures. The label asks prescribers to check valproate levels when these are started or stopped, so tell your prescriber before any change.
Choosing a product
What to look for on the label, and what to be skeptical of.
Look for…
Be skeptical of…
Frequently asked questions
Should I take L-carnitine with valproate?⌄
Only if your prescriber advises it. Valproate can lower carnitine and raise ammonia, and carnitine has lowered ammonia in case reports, but studies show low carnitine is not universal in long-term users. If you feel unusually tired, confused or sick to your stomach, ask your doctor whether to check your ammonia and carnitine levels.
Can I take aspirin with valproate?⌄
Talk to your prescriber or pharmacist first. Aspirin raised the active (free) part of valproate 4-fold in a small study, which the usual blood test may miss, and both can affect bleeding. The label advises caution. If a doctor prescribed aspirin for your heart, do not stop it without asking them.
Is CBD safe with valproate?⌄
Talk to your neurologist before using it. In trials of prescription CBD, liver enzyme rises were far more common in people also taking valproate (21% versus 3% without valproate or clobazam), and high ammonia has been reported. Liver tests may be needed.
Will biotin help hair loss from valproate?⌄
The evidence is thin. In one uncontrolled study in children, hair loss and rash improved with biotin, and valproate lowered the activity of the enzyme that recycles biotin. Ask your prescriber before trying it, and tell the lab if you take high-dose biotin, since it can distort some blood tests.
Can I drink alcohol on valproate?⌄
The Medication Guide says not to drink alcohol while taking valproate, because the two can add up to cause sleepiness and dizziness. Talk to your prescriber if this is a concern.
What if I could become pregnant?⌄
The label says valproate should be used in women who could become pregnant only when other medicines have failed or are unacceptable, with effective contraception, and never for migraine prevention in pregnancy. Folic acid is routinely recommended. If you become pregnant, do not stop suddenly; contact your prescriber right away to plan next steps.
References by claim
generalized and hard-to-classify epilepsy
childhood absence epilepsy
Glauser et al., 2010 · PMC (2010) link
acute manic episodes of bipolar disorder
Bowden et al., 1994 · PubMed (1994) link
preventing episodic migraine in adults
Linde et al., 2013 · PMC (2013) link
Safety
Depakote (divalproex sodium delayed-release tablets) prescribing information · DailyMed (FDA label) link
Interaction: Carnitine (L-carnitine)
Interaction: CBD (cannabidiol)
Epidiolex (cannabidiol) prescribing information · DailyMed (FDA label) link
Track Valproate (valproic acid, divalproex) with Pilora
Set up dose reminders, check interactions, and join the community in the Pilora app.
Disclaimer: This page summarizes published clinical-trial data for educational purposes and is not medical advice or a recommendation to use this prescription medication. Dosing, eligibility, and monitoring must be decided by a licensed prescriber.
