Supplement
Bitter Melon (Momordica charantia)
Glycemic measures in type 2 diabetes and prediabetes; pooled analyses disagree, the one Cochrane review found no preparation tested twice, and the largest change-score meta-analysis A study that pools the numbers from several trials into one combined estimate. found no effect on fasting glucose.
Grade C: human trials exist, but they are small, mixed or not yet repeated. This is not a finding that it does not work. How to read a grade →
Bitter melon is a vegetable used in Asian cooking and sold as a pill for blood sugar. Reviews of studies in people with diabetes or prediabetes disagree. One careful review found that no product had been tested twice. The largest review found no effect on long-term blood sugar. So the evidence is weak and mixed.
- Dose studied
- 500 mg–2 g/day (common range, not from cited trial)
- Form graded
- Momordica charantia extract
- Cited PMIDs PubMed ID: the permanent number the US National Library of Medicine gives a published paper. Each one links to that paper's record on PubMed.
- PMID 32951763 →
A letter describes the human evidence, not the product, the manufacturer, or the ingredient's worth.
See the full evidence record →Who responds: this record's trial reading reports no subgroup that identifies who responds.
Bitter melon (Momordica charantia) for glycemic markers, graded on human trial evidence.
How this letter has changed · latest: October 2026 · Corrected
The ref said Kim 2020 (PMID 32951763) 'modestly affected glycemic measures'; HbA1c A blood test that reflects average blood sugar over the previous two to three months. did not change and the fasting glucose fall is within-group only. The pooled analyses that show the inconsistency are now cited: Peter 2019 (PMID 30385422) and Zhang 2024 (PMID 38784554) positive, Laczko-Zold 2023 (PMID 38274207) and the Cochrane review Ooi 2012 (PMID 22895968) null, with the Cochrane finding that no preparation was tested twice. Mes 2025 (PMID 40199408) added for prediabetes. The record now says no single form or dose carries the letter. Letter unchanged.PMID 32951763 PMID 30385422 PMID 38784554 PMID 38274207 PMID 22895968 PMID 40199408
Every change to a letter on this site is dated and published. The full evidence log →
Commercial disclosure
EBL earns no commission on any supplement it grades. A purchase made through the dispensary link below pays EBL nothing, the practitioner discount goes to you in full. The dispensary is Fullscript, a practitioner dispensary, and many of the professional-grade brands in it are sold mainly through practitioners rather than in health food stores. Grades are assigned by published rubric before any commercial relationship exists and are never influenced by one. Full disclosure.
No studied form or dose carries the grade: trials used fresh fruit powder, juice, ethanol extracts and peptide-standardized extracts at 2 to 6 g/day of fruit material or 500 mg to 2 g/day of extract, over 4 to 16 weeks.
Standardized extracts/forms were used in the cited trials; product quality varies.
The summary of the cited 12-week trial does not state a time of day or whether doses were taken with food (PMID 32951763).
Checked against the published summary; the full paper could not be read. What the cited trials did, not a schedule. Every supplement, by trial timing →
Pregnancy: bitter melon has abortifacient activity in animal work. G6PD deficiency, on the vicine content of the seeds.
Diabetes medication, including insulin and sulfonylureas. Cochrane's review of bitter melon in type 2 diabetes found the evidence for glucose lowering insufficient rather than absent, so the additive-hypoglycemia caution stands on a weaker footing than for fenugreek or chromium, and it stands.
Bitter melon's glucose evidence is insufficient, not absent. Cochrane found the evidence for bitter melon lowering glucose in type 2 diabetes insufficient, which is why the additive-hypoglycemia caution here is weaker than for fenugreek or chromium without disappearing. Ooi et al. (Cochrane Database of Systematic Reviews A review that searches for every study on a question using a method set in advance, then assesses them together., 2012; PMID 22895968): Momordica charantia for type 2 diabetes mellitus.
Fasting blood glucose and HbA1c at baseline and 12 weeks. Postprandial glucose with a CGM during the first 4 weeks to characterize the individual response. Standardization varies widely: the most studied preparations used juice or extract standardized to charantin content. Hypoglycemia risk is real; monitored if taken with other glucose-lowering agents.
Primary evidenceSix papers, each linked to its PubMed record.
Kim et al. (Complement Ther Med, 2020; PMID 32951763): 90 adults with type 2 diabetes, extract for 12 weeks; HbA1c did not change in either group, and the fasting glucose fall (P = 0.014) is reported within the bitter melon group with no between-group difference against placebo An inactive substitute made to look like the real treatment, so participants cannot tell which one they received. stated. Pooled, positive: Peter et al. (J Ethnopharmacol, 2019; PMID 30385422), 10 trials, 1,045 patients, single-herb preparations lowered fasting glucose 0.72 mmol/L, postprandial glucose and HbA1c (0.26 percent) versus placebo, evidence rated low to very low quality; Zhang et al. (Heliyon, 2024; PMID 38784554), 8 trials, 423 patients, fasting glucose 0.85 mmol/L and HbA1c 0.38 percent lower, heterogeneity How much the results of the pooled trials differ from each other. High heterogeneity makes a combined estimate less reliable. high. Pooled, null: Laczko-Zold et al. (Front Nutr, 2023; PMID 38274207), 9 trials, 414 patients, no effect on fasting glucose (0.03 mmol/L, 95% CI -0.38 to 0.31) or HbA1c in change-score analysis; Ooi et al. (Cochrane, 2012; PMID 22895968), 4 trials, 479 participants, no significant difference from placebo, metformin or glibenclamide, risk of bias high, no preparation tested twice. Prediabetes: Mes et al. (J Ethnopharmacol, 2025; PMID 40199408), a 4-week crossover A trial design in which each participant receives both the treatment and the comparison, one after the other, and serves as their own control. (30 adults) null, and a 12-week parallel trial (38 adults) with lower fasting glucose, insulin and HOMA-IR on freeze-dried whole fruit. Grade C: randomized trials and pooled analyses that disagree, on surrogate measures, across preparations that are not interchangeable.
Educational use only · Not medical advice · Starting, stopping or changing any supplement or medication is a decision for your physician
These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease. The grade on this page describes the strength of published human trial evidence in the populations studied. It is not a recommendation for any individual and is not medical advice. Starting, stopping or changing any supplement or medication is a decision for your physician.
This page carries a letter. EBL takes no commission, referral fee or payment of any kind on anything it grades. No letter has been bought and none is for sale.
How this is enforced
That is not a promise, it is a build condition. Two checks run on every deploy: one refuses to publish if a commission is claimed on anything graded, the other refuses if a page names a commercial relationship that is not declared in the register. If either fails, this page does not go live. You can read how the checks work, and what they do not cover, on the evidence integrity page.
How EBL searched for the trials (1 search)
The search that was run, what it returned, and what was left out and why, so the absence of evidence can be checked as well as its presence.
(Momordica charantia[tiab] OR bitter melon[tiab] OR bitter gourd[tiab]) AND (glyc*[tiab] OR diabet*[tiab]) AND (randomized controlled trial[pt] OR meta-analysis[pt] OR systematic review[pt]) AND humans[mesh]Grades move when new trials publish. A free monthly email lists every grade EBL moved, with the reason and the trial behind it: the sign-up is on the Evidence Log →