Bitter melon and blood sugar: two meta-analyses that disagree

Bitter melon is sold as a natural way to lower blood sugar, and the research splits down the middle. A 2025 meta-analysis of randomized trials in people with prediabetes or type 2 diabetes found it lowered HbA1c by a standardized mean difference of 0.57 (95% CI 0.31 to 0.83). Its own authors graded that evidence very low, citing risk of bias, indirectness and publication bias, and the trials, which used different preparations, disagreed widely (I2 87 to 89%). A 2024 meta-analysis of placebo-controlled trials, working from change scores in 3 trials with 121 patients, found a difference of 0.12 points with a confidence interval crossing zero. A Cochrane review of 3 trials with 350 people with type 2 diabetes found no significant difference from placebo. Almost all of this tested powders, juices and extracts. As a cooked vegetable, a 130 g cup gives about 41.5 mg of vitamin C and 402 mg of potassium for 53 calories.

As people eat it, per 3.5 oz (100 g)

How much is actually in it

• USDA lists cooked bitter melon at 41 calories, 31.9 mg of vitamin C, 309 mg of potassium, 49 ug DFE of folate and 1.9 g of fiber per 3.5 oz (100 g). — national food composition database entry Evidence E source
  Per 100 g: 41 calories, vitamin C 31.9 mg, potassium 309 mg, folate 49 ug DFE, fiber 1.9 g

What a real portion looks like

• A cup of cooked bitter melon weighs 4.6 oz (130 g) and gives about 41.5 mg of vitamin C and 402 mg of potassium for 53 calories. — US dietary survey portion Evidence E source
  1 cup 130 g: vitamin C 41.5 mg, potassium 401.7 mg, 53.3 calories; 1 whole 120 g: vitamin C 38.3 mg

What it does to your health

• A 2025 meta-analysis of randomized trials found bitter melon lowered HbA1c by a standardized mean difference of 0.57 in prediabetes and type 2 diabetes, with very low certainty of evidence. — people with prediabetes or type 2 diabetes in 20 trials with 27 substudies for HbA1c (n = 20 trials) Evidence A source
  HbA1c SMD -0.57 (95% CI -0.83 to -0.31); fasting glucose SMD -0.46 (-0.78 to -0.14); I2 87-89%
• Freeze-dried whole bitter gourd at 3.6 g a day for 12 weeks lowered fasting glucose by about 0.9 mg/dL (0.05 mmol/L) per week in prediabetic adults, while a 4-week trial showed no difference. — prediabetic adults: 4-week crossover (n = 30, 2.4 g/day juice) and 12-week parallel trial (n = 38, 3.6 g/day whole fruit) (n = 38) Evidence B source
  FPG about -0.9 mg/dL (-0.05 mmol/L) per week vs unchanged on cucumber control; between-group p = 0.026
• In 75 adults with prediabetes, bitter melon extract at 300 or 600 mg a day for 12 weeks limited the rise in blood glucose to 1-3% against 11% on control. — 75 adults with prediabetes, 300 mg/day (n = 26), 600 mg/day (n = 24) or control (n = 25), 12 weeks (n = 75) Evidence B source
  Glucose change +1.05% (600 mg) and +3.35% (300 mg) vs +11.0% control, p < 0.05
• In 9 people with non-insulin-dependent diabetes, fried karela eaten daily produced a small improvement in glucose tolerance without higher insulin. — 9 non-insulin-dependent diabetics (plus rats) (n = 9) Evidence C source
  Small but significant improvement in glucose tolerance; water extract lowered glucose during a 50 g OGTT

Safety, limits and interactions

• Across pooled trials bitter melon caused no serious adverse effects, but 6 g of pulp a day gave significantly more diarrhea and flatulence than placebo. — participants of placebo-controlled bitter melon trials; liver and kidney markers from four trials (n = 4 trials) Evidence A source
  No significant change in ALT, AST or creatinine vs placebo; more diarrhea and flatulence at 6 g pulp/day
• A 60-year-old woman developed biopsy-confirmed acute interstitial nephritis after taking a pure Momordica charantia extract for 7 months, and recovered after stopping it. — one 60-year-old Korean woman (n = 1) Evidence D source
  Serum creatinine 1.45 mg/dL on admission, almost back to baseline 3 months after stopping and prednisolone

What people believe that the data does not support

• The EU did not authorize the claim that Momordica charantia helps to maintain a normal blood glucose. — EU health claim applicants Evidence E source
  Claim status: non-authorized

Who this works differently for

• In prediabetic adults in Tanzania, 2.5 g of bitter gourd powder daily for 8 weeks lowered fasting glucose by 5.6 mg/dL (0.31 mmol/L) more than cucumber juice. — 52 people with prediabetes, single-blinded crossover, 8 weeks per period (n = 52) Evidence B source
  FPG difference 5.6 mg/dL (95% CI 1.4 to 9.7; metric: 0.31 mmol/L, 0.08 to 0.54); larger effect at higher baseline glucose
• Among 1000 Sri Lankan outpatients with diabetes, non-prescribed native foods explained hypoglycemia in 16.9% of those affected, and bitter melon was involved in 54.5% of them. — 1000 patients with type 2 diabetes at a private diabetic clinic in Sri Lanka (n = 1000) Evidence C source
  26.1% had symptomatic hypoglycemia in the past month; native foods implicated in 16.9%, Momordica charantia in 54.5% of those
• In patients already on metformin plus dapagliflozin, adding a Momordica charantia extract for 3 months was associated with HbA1c falling from 6.93% to 6.34%. — 155 adults with type 2 diabetes on metformin; 70 with inadequate control on added dapagliflozin received the extract (n = 155) Evidence C source
  HbA1c 6.93% to 6.34% between months 6 and 9 in the extract group; greater fall than controls, p < 0.001

What is still unknown

• A 2024 meta-analysis of placebo-controlled trials found no HbA1c benefit of bitter melon on change scores, a mean difference of -0.12 points with a confidence interval crossing zero. — 3 trials with 121 patients for HbA1c change scores; 9 trials, 414 patients overall, 4-16 weeks (n = 121) Evidence A source
  HbA1c MD -0.12 (95% CI -0.35 to 0.11); post-intervention values borderline MD -0.24 (-0.49 to 0.00)
• A Cochrane review of 3 randomized trials with 350 participants found no significant glycemic difference between bitter melon preparations and placebo in type 2 diabetes. — 350 people with type 2 diabetes in 3 RCTs lasting up to 3 months (n = 350) Evidence A source
  No statistically significant difference vs placebo; leaf preparation comparable to glibenclamide in one trial
• A single dose of freeze-dried bitter melon at 50 or 100 mg/kg did not change glucose or insulin during a glucose tolerance test in 5 overweight men. — 5 healthy overweight non-diabetic men, three randomized conditions before a 75 g OGTT (n = 5) Evidence B source
  No significant difference in plasma glucose, insulin, energy expenditure or appetite

The bottom line

The trials that did find something were small and in people who already had high blood sugar. In 52 people with prediabetes in Tanzania, 2.5 g of bitter gourd powder a day for 8 weeks lowered fasting glucose by 5.6 mg/dL (0.31 mmol/L) more than cucumber juice, and not everyone responded. In 38 prediabetic adults, 3.6 g of freeze-dried whole fruit a day for 12 weeks lowered fasting glucose by about 0.9 mg/dL (0.05 mmol/L) per week, while a 4-week trial with juice showed no difference. In 75 adults with prediabetes, a commercial extract at 300 or 600 mg a day kept the 12-week rise in glucose to 1 to 3%, against 11% on control. In 5 overweight men without diabetes, a single dose changed nothing during a glucose test. The only study of the food as people cook it dates from 1981. In 9 people with diabetes, fried karela eaten daily gave a small improvement in glucose tolerance, in a trial without randomization. The EU refused the claim that it helps maintain normal blood glucose, and a refusal is not proof that it does nothing. Who should be careful. If you take diabetes medicine, the risk is too low a sugar. Among 1000 outpatients with diabetes in Sri Lanka, native foods taken without a prescription were blamed for hypoglycemia in 16.9% of those who had it, and bitter melon was involved in 54.5% of them, by self-report and without proof of cause. In 155 adults on metformin, those given an extract on top of dapagliflozin saw HbA1c fall from 6.93% to 6.34% over 3 months, in a study that did not randomize. Pooled trials found no serious harm and no change in liver enzymes or creatinine, with too little power to be sure, and 6 g of pulp a day caused more diarrhea and wind than placebo. One 60-year-old woman developed kidney inflammation after 7 months of a pure extract and recovered after stopping it. We found no human study in the corpus on pregnancy or children.

Sources

1. btm-r4-14 · national food composition dataset
   FNDDS 2709878 Bitter melon, cooked (75208310): Energy 41 KCAL/100 g · Vitamin C 31.9 MG · Potassium 309 MG · Folate, DFE 49 UG · Fiber 1.9 G
   USDA FNDDS 2021-2023 (FoodData Central), Bitter melon, cooked, food code 75208310, fdc_id 2709878 · checked 2026-09-29
2. btm-r4-15 · national food composition dataset
   FNDDS 2709878 Bitter melon, cooked: 1 cup 130 g · 1 whole 120 g · Vitamin C 41.5 MG/cup · Potassium 401.7 MG/cup · Energy 53.3 KCAL/cup
   USDA FNDDS 2021-2023 (FoodData Central), Bitter melon, cooked, food code 75208310, fdc_id 2709878 · checked 2026-09-29
3. btm-r4-01 · meta-analysis of RCTs
   The evidence showed that M. charantia, compared to control, significantly reduced HbA1c, SMD = −0.57, 95 %CI (−0,83, −0.31), p < 0.0001 ... For HbA1c, the risk of bias, indirectness, and publication bias were noted; therefore, the evidence was classified as very low.
   Metabolism Open, 2025 · checked 2026-09-29
4. btm-r4-06 · randomized controlled trial
   On average, bitter gourd intervention reduced FPG by ∼0.05 mmol/L per week, whereas FPG remained unchanged following placebo.
   J Ethnopharmacol, 2025 · checked 2026-09-29
5. btm-r4-07 · randomized controlled trial
   The high-dose (1.05% ± 10.2%) and low-dose bitter melon (3.35% ± 13.2%) groups showed smaller increases in blood glucose levels at 12 wk, compared to the control group (11.0% ± 16.3%) (p < 0.05).
   J Am Nutr Assoc, 2025 · checked 2026-09-29
6. btm-r4-12 · controlled clinical trial
   Fried karela fruits consumed as a daily supplement to the diet produced a small but significant improvement in glucose tolerance.
   Br Med J (Clin Res Ed), 1981 · checked 2026-09-29
7. btm-r4-03 · meta-analysis of RCTs
   The consumption of 6 g bitter melon pulp per day resulted in a significantly higher frequency of diarrhea and flatulence than in the placebo group (33). Based on data from four trials (33, 67, 68, 72), M. charantia administration exerted no significant effects on liver enzymes (ALT, AST) and creatinine levels (Supplementary Figure S6) compared to placebo, however confidence intervals were sometimes wide to draw reliable safety conclusions (i.e., power was low).
   Front Nutr, 2024 · checked 2026-09-29
8. btm-r4-11 · case report
   We report an extremely rare case of AIN that occurred after a patient ingested a pure M charantia extract over 7 months.
   Medicine (Baltimore), 2021 · checked 2026-09-29
9. btm-r4-13 · regulatory decision
   POL-HC-7601 Momordica charantia (Common Name : balsam pear, bitter melon, bitter gourd, ampalaya, karela (karola), fu kwa) Helps to maintain a normal blood glucose Non-authorised
   EU Register on nutrition and health claims, claim POL-HC-7601, snapshot 2026-09-21 · checked 2026-09-29
10. btm-r4-05 · randomized crossover trial
    The effect was also proven by a general linear mixed model (estimate 0.31; SE: 0.12; p: 0.01; 95%CI: 0.08; 0.54). Not all participants responded, but the higher the initial blood glucose levels were, the more pronounced the effect was.
    J Ethnopharmacol, 2018 · checked 2026-09-29
11. btm-r4-09 · observational study
    Non-prescribed native food items accounted for hypoglycaemia in 16.9% of patients (Momordica charantia 54.5%, Costus speciosus 52.3%, Salacia prinoides 11.4%, Coccinia grandis 6.8%, Adenanthera pavonina 4.5%).
    BMC Endocr Disord, 2018 · checked 2026-09-29
12. btm-r4-10 · observational cohort
    In the intervention group, HbA1c decreased from 7.82 ± 0.58% at baseline to 6.93 ± 0.30% at 6 months and to 6.34 ± 0.42% at 9 months, while in the control group, glycemic parameters showed only modest additional changes after 6 months.
    Nutrients, 2026 · checked 2026-09-29
13. btm-r4-02 · meta-analysis of RCTs
    The meta-analysis of the post-intervention values using the random effects model indicated borderline efficacy for the M. charantia treatment with an MD = −0.24 (95% CI, −0.49 to 0.00, I2 = 0%) (Figure 3B), however, the analysis of the changes scores definitely did not show superiority over placebo (MD = −0.12; 95% CI: −0.35 to 0.11; I2 = 56%) (Figure 3A).
    Front Nutr, 2024 · checked 2026-09-29
14. btm-r4-04 · systematic review of RCTs
    Two RCTs compared the effect of preparations from different parts of the momordica charantia plants and placebo on the glycemic control in type 2 diabetes mellitus. There was no statistically significant difference compared to placebo. ... There is insufficient evidence to recommend momordica charantia for type 2 diabetes mellitus.
    Cochrane Database Syst Rev, 2010 · checked 2026-09-29
15. btm-r4-08 · randomized crossover trial
    Plasma glucose and insulin levels significantly increased during the OGTT (p < or =0.05) but no significant difference was observed between experimental conditions.
    J Ethnopharmacol, 2009 · checked 2026-09-29