The palmitoylethanolamide evidence base — including the parts that did not work.
Seven meta-analyses, dozens of randomized trials, a safety database of roughly 1,590 patients, and a set of null results that any serious buyer should see before they commit. This page is the whole picture, not the flattering half of it.
Key human and animal trials at a glance
Filter by category. Negative and null findings are included by design — they are marked, not buried.
| Study | Design | N | Dose | Duration | Principal finding |
|---|---|---|---|---|---|
| Viña & López-Moreno, 2025 Nutrition Reviews | Meta-analysis, 18 RCTs | 1,196 | 300–1200 mg/d | 6–26 wk | SMD −0.90 at 6 wk deepening to −1.16 at 24–26 wk. Effective across nociceptive (−0.74), neuropathic (−0.97) and nociplastic (−0.59) pain. QoL SMD −0.61. |
| Lang-Illievich et al., 2023 Nutrients | Meta-analysis, 11 double-blind RCTs | 774 | 300–1200 mg/d | — | Pooled SMD 1.68 (95% CI 1.05–2.31, p = 0.00001) for pain intensity. Quality of life, function and sleep also improved. Side effects negligible. |
| Schweiger / Polati et al., 2024 Nutrients | Meta-analysis, 9 studies | 742 | 600–1800 mg/d | 60 d | A further 1.36-point pain reduction accrued between day 30 and day 60 — evidence that a 30-day trial under-reads the effect. Cumulative reduction 3.44 points by day 60. |
| Paladini et al., 2016 Pain Physician | Pooled analysis, 12 studies | ~1,188 | — | 60 d | Pain fell 1.04 points per two weeks on PEA vs 0.20 on control. 81% of PEA patients reached pain ≤3/10 by day 60 vs 40.9% of controls. |
| Scuteri et al., 2022 Pharmaceutics | Systematic review & meta-analysis, 10 studies | 933 | — | — | Significant benefit overall (p<0.00001) — but the two randomized trials alone did not reach significance. Heterogeneity I² = 99%. Authors call for adequately powered RCTs. |
| Sciatic nerve compression RCT Via J Pain Res review | Randomized, placebo-controlled | 636 | 300 or 600 mg/d | 3 wk | VAS fell 7.1 → 2.1 in the 600 mg arm. NNT of 1.5 for 50% pain reduction — among the strongest numbers in the whole PEA literature. |
| Pickering et al., 2022 Inflammopharmacology | Quadruple-blind RCT | 66 | 600 mg/d | 8 wk | Diabetic peripheral neuropathic pain. Significant reduction in BPI-DPN severity and interference (p<0.001) and NPSI total (p≤0.001). Sleep and depression scores improved; IL-6 fell (p = 0.04). |
| Lang-Illievich et al., 2022 Nutrients | Randomized crossover, healthy volunteers | 14 | 1200 mg/d | 4 wk/arm | Improved temporal summation, conditioned pain modulation, and cold and pressure pain tolerance — i.e. measurable effects on central and peripheral sensitization. No adverse events. |
| Steels et al., 2019 Inflammopharmacology | Double-blind, placebo-controlled, 3 arms | 111 | 300 or 600 mg/d | 8 wk | Knee osteoarthritis. WOMAC total reduced at both doses (p = 0.037 and p = 0.0012); pain subscale p = 0.0074 and p<0.001. Anxiety scores also improved. Well tolerated. |
| Del Giorno et al., 2015 Pain and Therapy | Prospective & retrospective observational | 80 | 1200 → 600 mg/d | 3 mo | Fibromyalgia, added to duloxetine and pregabalin. Tender points fell from 14/18 to 1/18 (p<0.0001). Not placebo-controlled — observational. |
| Rao et al., 2021 Sleep Science & Practice | Double-blind, randomized, placebo-controlled | 103 | 350 mg pre-bed | 8 wk | Sleep. Significant reduction in sleep-onset latency in the subgroup with baseline latency >10 min at weeks 4 and 8; improved wake-up cognition. Overall PSQI not significantly different between groups. |
| Kim et al., 2024 Nutrients | Randomized crossover | 39 | 700 mg/d | 6 wk/arm | Healthy young adults. Significant increase in serum BDNF; improved memory accuracy and fewer errors on the CANTAB paired-associates task versus placebo. |
| Piccolo et al., 2025 Frontiers in Nutrition | Randomized clinical trial | 60 | 1200 mg + 0.2 mg melatonin | 3 mo | Migraine prevention. Migraine days 3.4 → 2.2 (p<0.001); attack duration 10.0 → 7.1 h; moderate-to-severe disability 93% → 30%. Combination product, not PEA alone. |
| Indraccolo et al., 2017 Ann Ist Super Sanità | Meta-analysis, 4 studies | — | 400 mg PEA + 40 mg polydatin BID | 3 mo | Endometriosis-related pain. Clinically relevant improvement in chronic pelvic pain and dysmenorrhoea; limited effect on deep dyspareunia; none on dyschezia. Authors rate the underlying studies as poor quality. |
| Levagen+ canine & feline trial, 2026 Front Vet Sci | Double-blind, placebo-controlled RCT | 50 dogs, 50 cats | — | 6 wk | Joint pain. 76% of dogs classed as successfully treated vs 40% on placebo. Cats showed significant improvement in jumping and pain at weeks 2 and 6. Well tolerated in both species. |
| Andresen et al., 2016 Pain | Multicenter randomized, double-blind, placebo-controlled | 73 | Ultra-micronized PEA | 12 wk | Negative trial. Spinal cord injury neuropathic pain. No significant difference in pain intensity (p = 0.46); PEA 0.4-point reduction vs placebo 0.7. No effect on spasticity, insomnia or psychological function. No excess adverse events. |
| Faig-Martí & Martínez-Catassús, 2017 J Orthop Traumatol | Prospective randomized | 61 | 600 mg/d | 60 d | Largely negative. Carpal tunnel syndrome. No significant improvement in Durkan's or Phalen's test, VAS, or electrophysiology. Only the Boston Questionnaire improved. Authors concluded the dose was insufficient for this population. |
| Plesnik, 1977 Impulsin program | Double-blind placebo-controlled | 393–457 children | 600 mg/d | 8 wk | Not statistically significant. Respiratory infection prophylaxis. 15.7% fewer acute respiratory infections overall — attributed by the authors to short treatment duration and the absence of an influenza epidemic during the study window. |
Summarized from published abstracts and full texts. Provided for scientific and technical reference to trade and professional audiences. Not a claim that any PEAOID product produces these outcomes.
Four caveats every buyer should be able to answer
Your medical-affairs reviewer will raise these. Better that they come from your supplier first.
Heterogeneity is very high
I² values of 95–99% run through the pooled pain analyses. The trials differ in population, dose, formulation, duration and outcome measure. The direction of effect is consistent; the magnitude should be read as a range, not a number.
Geographic concentration
A large share of the positive literature originates from Italian research groups, much of it funded by or affiliated with the company that originally commercialised PEA in Europe. That does not make it wrong. It does mean independent replication outside Italy carries extra weight.
Open-label studies do heavy lifting
In the 2022 systematic review, restricting the analysis to the randomized trials alone caused the pooled effect to lose statistical significance. The double-blind subset is real and positive — the 2023 meta-analysis of eleven double-blind RCTs is the strongest single piece of evidence — but the open-label data inflates the headline picture.
Long-term data is thin
Most trials run 8 to 12 weeks. The pooled safety conclusion that excludes serious adverse reactions at 1 in 200 was explicitly limited to courses of up to 49 days. There is no controlled multi-year human safety dataset.
Consistently, unusually well tolerated.
Across the entire reviewed literature, PEA's adverse-event burden is comparable to or lower than placebo. This is the strongest single argument for the ingredient commercially.
- NOAEL above 1,000 mg/kg/day in a GLP-compliant OECD TG414 rat developmental toxicity study — no adverse findings at the highest dose tested.
- Serious adverse reactions excluded at ≥1 in 200 for treatment courses up to 49 days, across roughly 1,590 patients in 16 trials.
- Dropout lower on PEA than on placebo in the pooled randomized data — approximately 1.1% versus 4.3%.
- Reported events are mild: occasional gastrointestinal discomfort, transient drowsiness, headache, isolated palpitations.
- Used alongside tramadol, pregabalin, gabapentin and duloxetine in trials without reported adverse interaction — though no formal interaction study exists.
Gaps in the safety picture
Pregnancy and lactation. No human data. Precautionary labeling is appropriate.
Beyond 60 days. The pooled analysis explicitly declines to extend its conclusion. Rarer events at frequencies below 1 in 100 cannot be ruled out for longer courses.
Drug interactions. No published pharmacokinetic interaction studies. A theoretical mechanism-based interaction with fibrate PPAR-α agonists has been raised but not demonstrated.
Pediatric use. Limited. The largest pediatric dataset is the 1977 Impulsin respiratory trial in children aged 11–15, which was well tolerated but did not meet its efficacy endpoint.
What doses the trials actually used
| Daily dose | Regimen | Studied in |
|---|---|---|
| 300 mg | Once or twice daily | Knee osteoarthritis (low arm), sciatica (low arm) |
| 350 mg | Single dose before bed | Sleep-onset latency |
| 600 mg | 300 mg twice daily | Sciatica, diabetic neuropathy, knee osteoarthritis, carpal tunnel |
| 700 mg | Daily | Cognition and BDNF, healthy adults |
| 1,200 mg | 600 mg twice daily | Fibromyalgia loading month, migraine prevention, sensitization studies |
| Up to 1,800 mg | Divided | Historical Impulsin respiratory program; reported as well tolerated |
The most useful practical finding
The 2024 extended-treatment meta-analysis found roughly 60% of the total pain reduction occurring in month one, with a further 1.36-point reduction accruing between day 30 and day 60. The implication is methodological: any evaluation of PEA that runs only 30 days — a clinical study, a consumer trial or an internal assessment — is reading the ingredient at well under its observed effect. What a finished product may say about that is a matter for the marketer and the local regulator, not for us.
Want the primary papers rather than our summary?
We will send the reference list with PMIDs and DOIs, the full-text PDFs we are permitted to share, and a written position on any claim you are considering — including the ones we think you should not make.