Low-Dose Naltrexone (LDN): Off-Label Doses, Evidence by Condition and Side Effects

LDN is a different use from 50 mg naltrexone

Naltrexone is approved to treat alcohol dependence and to block the effects of opioids, at 50 mg once a day for most people.¹⁰ LDN uses roughly a tenth of that dose or less, for conditions the FDA has not approved it for.^2,9 Contrave, a weight-loss tablet, also contains naltrexone, but it is a different approved product that pairs 8 mg of naltrexone with bupropion in each tablet.¹¹

Contrave's label says taking it with other naltrexone products is not recommended.¹¹ Label doses for the 50 mg tablet and the Vivitrol shot are on naltrexone dosage.

No LDN product is approved, so it is compounded

In the FDA's National Drug Code directory, every single-ingredient oral naltrexone listing is a 50 mg film-coated tablet. The directory also lists naltrexone powder as a bulk ingredient for prescription compounding.¹ A 2026 evidence review describes LDN supply as limited mainly to compounding pharmacies, at the prescriber's discretion.²

That has three consequences for anyone considering it:

• It needs a prescription. Naltrexone is a prescription-only drug, so LDN is not sold over the counter.¹⁰ It is not a controlled substance under federal law.¹² See is naltrexone a controlled substance?
• No FDA quality check. "Compounded drugs are not FDA-approved," and the FDA does not verify their safety, effectiveness or quality before they are marketed.⁷ The 2026 review lists inconsistent compounded formulations as one source of uncertainty in LDN research.²
• Home-made doses vary. Some people cut 50 mg tablets or dissolve them to make their own low doses. Researchers who ran early LDN trials warned that this likely causes day-to-day variation in the dose.⁹

Cost: about $25 to $99 a month (October 2026 listings)

On October 17, 2026, online LDN sellers advertised about $25 to $99 for a month's supply, without insurance.^13,14 The $25 figure is a first-quarter promotion. These are advertised prices, not checked at checkout, and promotions change.

In 2014, a review by LDN researchers estimated that LDN "appears to average US$35 per month," including compounding and without insurance, and noted that insurance plans may not cover it.⁹ Prices for the 50 mg tablet and the Vivitrol shot are on naltrexone cost.

None of the 15 alcohol and opioid treatment programs in our provider directory mention LDN. The sellers in the table do. A listing here is not a recommendation, and every use of LDN is off-label.^1,2

Doses in studies: mostly 4.5 mg a day

Most LDN studies gave 4.5 mg once a day, often reached by raising the dose in steps.² The 2026 review calls LDN dosing "empirically derived rather than evidence-based."²

A 2014 review by one trial group described LDN taken about an hour before bed, moved to the morning for people with insomnia, and lowered to 3 mg for people with side effects.⁹ Trials raise the dose on their own schedules. A registered Canadian long COVID trial, for example, starts at 1 mg and reaches 4.5 mg by week 4.²⁵ Your prescriber sets any dose.

Stopping. The same trial group saw no dependence or tolerance with LDN. After stopping, symptoms generally returned slowly to where they had been.⁹ Small studies observed this; no trial has tested it as an outcome.

Evidence by condition

Many LDN trials are small and short, and rely on symptoms people report themselves, which are vulnerable to placebo effects.² The table lists each controlled trial once. The two meta-analyses at the bottom pool some of the same fibromyalgia trials and were published before the 2026 INNOVA trial.

Fibromyalgia: early crossover studies were positive, larger recent trials were not

Two small early studies, one of them single-blind, found less pain on LDN.^21,26 Three later trials did not. In a Danish trial of 99 women, the pain difference was 0.34 points on a 0-10 scale (p=0.27).⁴ In the Spanish INNOVA trial of 98 women followed for a year, LDN "did not demonstrate a clinically meaningful benefit over placebo."⁵ A 2024 meta-analysis of 4 trials did find lower pain scores on LDN, but it predates INNOVA.²⁹ Pain and fibromyalgia will get their own page (see pages in this guide).

Crohn's disease: one small adult trial, low-quality evidence

The Cochrane review found two trials with 46 people in total. In the adult trial, more people on LDN had a clinical response (83% vs 38%), but remission rates did not differ significantly. The evidence was rated low quality, with too little data for "any firm conclusions."²²

Multiple sclerosis: two 2010 trials, mixed results

One crossover trial found better mental-health quality-of-life scores on LDN, but lost a quarter of its participants.²³ A second trial of 96 people found no meaningful differences, and its authors wrote that LDN's efficacy "is under question."²⁷

Long COVID: no randomized trial results yet

A 2026 systematic review found no randomized trials, only 4 before-and-after studies with 155 people in the US and Ireland, and rated the certainty of evidence as low.²⁵ Four registered trials are under way, 2 of them randomized and controlled.²⁵

How LDN might work is still a hypothesis

Three explanations have been proposed. LDN might block a receptor called TLR4 on microglia, immune cells in the brain and spinal cord, and so lower inflammation. A brief opioid block might make the body produce more of its own opioids, such as endorphins. Or it might act on a cell-growth receptor (the opioid growth factor receptor).^2,9 The 2026 review calls these mechanisms "largely theoretical and speculative," with limited high-quality clinical evidence that they matter in human disease.²

Side effects reported in trials

Fibromyalgia symptoms such as poor sleep and fatigue overlap with common side effects, which may make side effects harder to attribute in these trials.⁵ Studies designed to test LDN's safety are sparse, and long-term safety at low doses is largely unknown.^2,9 Rates trial by trial, serious adverse events, how long side effects last and what happens when people stop are on low-dose naltrexone side effects. Side effects of the 50 mg tablet are on naltrexone side effects.

Opioids: the interaction that matters

The 50 mg label says naltrexone must not be used by people taking opioid pain medicines or who depend on opioids, including methadone or buprenorphine.¹⁰ At full dose, naltrexone can set off sudden opioid withdrawal in someone with opioids in their body.¹⁰

For LDN, the evidence is thinner. Researchers who ran early trials wrote that even a low dose "could cause a sufficient blockade of opioid receptors to reduce the effectiveness of opioid analgesics." They excluded everyone taking opioid painkillers and, as of 2014, knew of no studies combining the two at LDN doses. They also knew of no reports of LDN interacting with other medicines, while noting that small trials leave many combinations untested.⁹ A 2017 case report adds a different risk, over-sedation after a single oxycodone dose on LDN; details are on what to avoid on LDN.³¹ Planning around surgery on standard naltrexone is covered in stopping naltrexone before surgery.

Surgery on LDN has only expert-opinion guidance: two Australian anaesthetists suggest holding it for 24 to 72 hours beforehand.³² If you take LDN, tell the surgeon and anesthesiologist, and agree on a stop date with them and your prescriber; the details are in surgery and procedures.

Kratom, opioid cough and diarrhea medicines, alcohol and the other items on common LDN avoid lists are graded by evidence, one by one, in what to avoid on low-dose naltrexone.

Who shouldn't take LDN

There is no LDN label. The 50 mg tablet label lists five groups who should not take naltrexone¹⁰:

1. People taking opioid pain medicines
2. People who depend on opioids, including those on methadone or buprenorphine
3. People in opioid withdrawal
4. Anyone who fails a naloxone challenge test or has a positive urine test for opioids
5. Anyone who has had a reaction to naltrexone or the tablet's ingredients

The same label says safe use under 18 has not been established and asks patients to tell their doctor if they are pregnant, plan to be, or are breastfeeding.¹⁰ Whether these cautions apply at low doses is unknown (pregnancy, liver and kidney disease on LDN).² The one naltrexone case in a 20-year US liver-injury registry was at 4.5 mg a day; it was mild and resolved within a week of stopping.³³ See naltrexone and your liver. The 50 mg label also warns that people who have used opioids may be more sensitive to them after stopping naltrexone (overdose warning: naltrexone side effects).¹⁰ Medicines, supplements and procedures to avoid or ask about are on what to avoid on LDN.

LDN has no evidence for alcohol use disorder

The alcohol trials behind naltrexone's label used 50 mg a day, and that is the label dose.¹⁰ The Sinclair method also uses the 50 mg tablet, taken before drinking.³⁴ See the Sinclair method and naltrexone for alcohol use disorder.

One small trial gave LDN to 45 people with HIV and chronic pain in St. Petersburg, Russia, 15 per group, who had drunk heavily in the past year; at the start they averaged about 2 heavy drinking days a month. Over 8 weeks, the share of heavy drinking days rose by 3.07 points on 4.5 mg LDN and fell by 4.63 points on placebo, a difference that was not statistically significant.⁶ Pain was the main outcome, and drinking a secondary measure in a pilot this size. The authors note that naltrexone "at higher doses (50mg or more)" is the established alcohol treatment.⁶

Pages in this guide

Related: LDN on how long naltrexone stays in your system · LDN on naltrexone dosage

Your experience: If you have taken LDN, you can share your story on our naltrexone reviews page. LDN stories will be labeled off-label.

How we checked

• Trial results come from the published trials and systematic reviews (Cochrane 2018, BMJ Open 2026, two pain meta-analyses), read on PubMed on October 17, 2026, with full texts where open.^22,25,29,30
• Statements about the whole evidence base come from a 2026 review of 105 studies.²
• Approval status comes from the FDA's drug directory and compounding guidance.^1,7
• Each trial is counted once.