Does magnesium lower blood sugar?

A little, and most clearly in people who already have diabetes. The largest pooled analysis, of 78 randomized trials in a mix of healthy and ill people, found magnesium supplements lowered fasting blood sugar by 3.6 mg/dL and HbA1c by 0.15 percentage points against controls. Its authors call these effects modest and of uncertain clinical relevance. They rated the HbA1c evidence moderate certainty and the fasting glucose evidence low.

Established. Magnesium supplements lower blood sugar measures by a small amount on average. A 2026 meta-analysis of 78 randomized trials found HbA1c down 0.15 points (95% CI 0.03 to 0.26) on moderate-certainty evidence. Most of those trials used 300 mg a day or more.

Unsettled. Whether that drop matters for a person's health is open. Two meta-analyses found HbA1c barely moved, and one older review found no significant change in glucose at all. No trial we found used diabetes itself as the outcome.

Drop in fasting blood sugar against control, mg/dL

Both rows are pooled differences from randomized trials, with 95% confidence intervals. The gestational diabetes interval is very narrow for five small trials, which suggests a fixed-effect model. Sources: cards ev-mgbs-01 and ev-mgbs-08 below.

What the trials found

The 2026 analysis of 78 trials is the broadest. Fasting blood sugar fell by 3.60 mg/dL, with a 95% confidence interval from 1.06 to 6.13. HbA1c fell by 0.15 percentage points. Insulin resistance, measured as HOMA-IR, fell by 0.48. The trials mixed healthy people with people who had diabetes or other conditions.

Trials limited to type 2 diabetes tell a similar story with a sharper edge. A 2025 meta-analysis of 23 trials in 1,345 people with type 2 diabetes found fasting blood sugar lower and called the change in HbA1c minimal, a difference of 0.16 with an interval from 0.32 down to zero. A 2016 analysis of double-blind trials found fasting glucose lower in people with diabetes, a standardized difference of 0.40 whose interval also reached zero, with high variation between trials.

Dose may matter. A 2022 dose-response meta-analysis of 18 trials in type 2 diabetes estimated HbA1c 0.73 points lower at 500 mg a day (95% CI 0.22 to 1.25). That is a modeled estimate, at a dose above the 350 mg upper limit for supplemental magnesium, and the authors say it is not enough for guidelines.

Some results point the other way. A 2016 meta-analysis of trials in people with and without diabetes found magnesium improved HOMA-IR by 0.67 but did not significantly change plasma glucose, HbA1c or insulin. A 2026 meta-analysis of 15 placebo-controlled trials in 1,085 people with diabetes or prediabetes found the drops in insulin and HOMA-IR did not reach statistical significance. In a 6-week crossover trial of 14 people with insulin-treated type 2 diabetes and low magnesium, insulin sensitivity measured with a clamp did not improve, although serum magnesium rose a little.

Pregnancy and prediabetes

In gestational diabetes, five randomized trials in 266 women found magnesium lowered fasting blood sugar by 7.33 mg/dL and HOMA-IR by 0.99. It made no clear difference to preterm birth or large babies. The trials were small and mostly from one country. They tested magnesium in women already being treated for gestational diabetes, and none of the sources here sets a separate supplement limit or safety profile for pregnancy. That gap is not evidence of safety.

In a 4-month trial of 71 older adults in China with prediabetes and low blood magnesium, 360 mg a day of magnesium oxide lowered fasting blood sugar by about 9 mg/dL (0.5 mmol/L) against placebo. The authors describe it as exploratory, and the groups differed in insulin at the start.

The population studies

Observational data from 25 cohort studies in 637,922 people link each extra 100 mg a day of magnesium from food with an 8 to 13% lower risk of developing type 2 diabetes. Those figures are relative, and the pooled analysis gives no absolute risk. People who eat more magnesium also eat more fiber and whole grains, so this cannot show that magnesium is the cause.

Who should be careful

Not for everyone. If your kidneys do not work well, magnesium supplements can build up. The NIH Office of Dietary Supplements warns that the risk of magnesium toxicity rises with impaired kidney function or kidney failure, because the body loses the ability to clear the excess. Kidney disease is a known complication of long-standing diabetes, so this applies to many of the people asking the question.

The upper limit for supplemental magnesium is 350 mg a day for adults and 65 to 350 mg for children and teens, depending on age. It does not count magnesium from food. Several of the blood sugar trials used doses near or above it, including the 500 mg estimate above. Magnesium carbonate, chloride, gluconate and oxide are the forms most often reported to cause diarrhea, and oxide was the form in the prediabetes trial. Magnesium can interact with oral bisphosphonates and with tetracycline and quinolone antibiotics. We found no card on magnesium taken alongside diabetes medicines, and that gap is not evidence of safety.

What expert bodies say

The NIH Office of Dietary Supplements reports that, according to the American Diabetes Association, there is not enough evidence to recommend magnesium to improve blood sugar control in people with diabetes. The ADA reference it cites predates the meta-analyses from 2022 to 2026. The same fact sheet notes that people with insulin resistance or type 2 diabetes can lose more magnesium in urine and run low, which may be why trials in people with low magnesium are the ones to watch.

How we searched

Searched: a local copy of PubMed, on 7 October 2026. The broadest query paired magnesium with glucose, blood sugar, HbA1c, insulin resistance or diabetes (1,288 hits, top 40 screened). A second covered dietary magnesium and diabetes risk in cohorts (108 hits, top 25 screened). A third covered prediabetes and insulin sensitivity (105 hits, top 25 screened).

Also searched: the NIH Office of Dietary Supplements fact sheet, ClinicalTrials.gov (2 trials, none with blood sugar as the main outcome and results posted), Retraction Watch for every cited paper (none retracted), the web for 2025 and 2026 meta-analyses and the ADA position, and Europe PMC for recent prediabetes trials.

Included: seven meta-analyses of randomized trials, two randomized trials, one meta-analysis of cohort studies and the NIH fact sheet. Seventeen cards in total: 8 at level A, 2 at B, 1 at C and 6 at E.

Excluded: a meta-analysis in polycystic ovary syndrome that pooled several minerals, a pooled analysis that overlaps the 2025 review, an older gestational diabetes review replaced by a newer one, a network meta-analysis with no magnesium-specific number, older cohort analyses replaced by a newer one, and news stories.

What we read: abstracts, each quotation checked against the abstract text, and the results section of the 2026 meta-analysis in full text for its certainty grades.

What we could not get: the full text of the 2025 type 2 diabetes meta-analysis, so the units of its glucose result are unknown. The ADA Standards of Care text itself, so its position is cited through the NIH fact sheet.

What would change this answer

A large randomized trial lasting six months or more in people with prediabetes and low magnesium, with HbA1c or new diabetes as the outcome. We found none registered.
Trials that report who responds. The pattern so far points to people with diabetes or low magnesium, and most trials did not select for either.
Safety data for magnesium taken with diabetes medicines and in diabetic kidney disease.

The rest of the nutrient, in one place. Magnesium: widely under-eaten, and widely over-promised → What it does, how much you need, who runs short, and what too much does, with the evidence level shown on every line.

The full guide

Magnesium: widely under-eaten, and widely over-promised

The same question for other foods (blood sugar)

Sources

ev-mgbs-01 · Meta-analysis or systematic review · systematic review and dose-response meta-analysis of randomized controlled trials · n = 78 trials
Mg supplementation significantly reduced body weight (weighted mean difference [WMD]: -0.70 kg; 95% confidence interval [CI]: -1.30, -0.09), diastolic blood pressure (WMD: -1.58 mmHg; 95% CI: -2.50, -0.65), homeostasis model assessment of insulin resistance (WMD: -0.48; 95% CI: -0.79, -0.18), low-density lipoprotein cholesterol (WMD: -3.21 mg/dL; 95% CI: -5.27, -1.15), fasting blood glucose (WMD: -3.60 mg/dL; 95% CI: -6.13, -1.06), systolic blood pressure (WMD: -2.50 mmHg; 95% CI: -4.08, -0.91), glycated hemoglobin (WMD: -0.15%; 95% CI: -0.26, -0.03), triglycerides (WMD: -9.24 mg/dL; 95% CI: -16.87, -1.61), and interleukin-6 levels (WMD: -1.10 pg/mL; 95% CI: -1.93, -0.27) compared with controls.
Who: participants in randomized controlled trials of magnesium supplementation, searched to May 2026
Effect: fasting blood glucose WMD -3.60 mg/dL (95% CI -6.13 to -1.06); HbA1c WMD -0.15% (-0.26 to -0.03); HOMA-IR WMD -0.48 (-0.79 to -0.18); fasting insulin not significant
Certainty: Mixed populations, healthy and ill; 79.5% of trials used 300 mg/day or more; the authors call the effects modest and of uncertain clinical relevance.
Nutrients, 2026 · checked 2026-10-07 · we read the abstract
ev-mgbs-02 · Meta-analysis or systematic review · systematic review and dose-response meta-analysis of randomized controlled trials, GRADE · n = 78 trials
Moderate-certainty evidence was observed for BW, BMI, HC, HbA1c, HOMA-IR, DBP, IL-6, CRP, and Cr.
Who: randomized controlled trials of magnesium supplementation (GRADE assessment per outcome)
Effect: GRADE: HbA1c and HOMA-IR moderate; fasting blood glucose and fasting insulin low
Certainty: The low rating for fasting glucose sits in the next sentence of the same section: 'Low-certainty evidence was found for WC, FBG, FI ...'.
Nutrients, 2026 · checked 2026-10-07 · we read the fulltext
ev-mgbs-03 · Meta-analysis or systematic review · systematic review and meta-analysis of randomized controlled trials · n = 1345
However, the impact on glycated haemoglobin was minimal (WMD = -0.16, 95% CI: -0.32 to 0.00).
Who: patients with type 2 diabetes in randomized controlled trials
Effect: fasting blood glucose WMD -0.58 (95% CI -0.87 to -0.28); HbA1c WMD -0.16 (-0.32 to 0.00); serum Mg WMD +0.69
Certainty: The abstract gives no units; the HbA1c interval touches zero. Larger HbA1c drop in people 65 and over and with longer supplementation.
Sultan Qaboos Univ Med J, 2025 · checked 2026-10-07 · we read the abstract
ev-mgbs-04 · Meta-analysis or systematic review · updated systematic review and meta-analysis of randomized controlled trials · n = 1085
Overall, the oral magnesium supplementation did not result in a statistically significant change in insulin levels (MD = - 1.73; 95% CI: -3.73 to 0.27; p-value = 0.09) and HOMA-IR (MD = - 0.74; 95% CI: -1.58 to 0.09; p-value = 0.08) in people with diabetes.
Who: individuals with diabetes and prediabetes in placebo-controlled RCTs, searched to April 2026
Effect: insulin MD -1.73 (95% CI -3.73 to 0.27, p=0.09); HOMA-IR MD -0.74 (-1.58 to 0.09, p=0.08); prediabetes also non-significant
Certainty: Point estimates favor magnesium; meta-regression suggests people with higher baseline insulin respond more.
BMC Nutr, 2026 · checked 2026-10-07 · we read the abstract
ev-mgbs-05 · Meta-analysis or systematic review · systematic review and meta-analysis of randomized controlled trials · n = 22 treatment arms
A significant effect of magnesium supplementation was observed on HOMA-IR index (WMD: -0.67, 95% CI: -1.20, -0.14, p=0.013) but not on plasma glucose (WMD: -0.20mmol/L, 95% CI: -0.45, 0.05, p=0.119), HbA1c (WMD: 0.018mmol/L, 95% CI: -0.10, 0.13, p=0.756), and insulin (WMD: -2.22mmol/L, 95% CI: -9.62, 5.17, p=0.556).
Who: diabetic and non-diabetic individuals in RCTs of oral magnesium (22 treatment arms for glucose, 14 for HbA1c)
Effect: HOMA-IR WMD -0.67 (95% CI -1.20 to -0.14); plasma glucose WMD -3.6 mg/dL (-8.1 to 0.9; metric: -0.20 mmol/L, -0.45 to 0.05), not significant; HbA1c not significant
Certainty: Duration subgroup (4 months or more) is post hoc.
Pharmacol Res, 2016 · checked 2026-10-07 · we read the abstract
ev-mgbs-06 · Meta-analysis or systematic review · systematic review and meta-analysis of double-blind randomized controlled trials · n = 670
Compared with placebo (n=334), Mg treatment (n=336) reduced fasting plasma glucose (studies=9; SMD=-0.40; 95% CI: -0.80 to -0.00; I2=77%) in people with diabetes.
Who: people with diabetes (12 trials) or at high risk of diabetes (6 trials), oral magnesium vs placebo
Effect: fasting plasma glucose in diabetes SMD -0.40 (95% CI -0.80 to -0.00), I2 77%
Certainty: Interval for diabetes reaches zero; high heterogeneity; standardized units only.
Eur J Clin Nutr, 2016 · checked 2026-10-07 · we read the abstract
ev-mgbs-07 · Meta-analysis or systematic review · systematic review and dose-response meta-analysis of controlled clinical trials · n = 18 trials
The dose-response testing indicated that the estimated mean difference in HbA1c at 500 mg/d was -0·73 % (95 % CI: -1·25, -0·22) suggesting modest improvement in HbA1c with strong evidence (P value: 0·004).
Who: patients with type 2 diabetes in randomized clinical trials of oral magnesium
Effect: HbA1c MD -0.73% at 500 mg/day (95% CI -1.25 to -0.22)
Certainty: 500 mg/day is above the 350 mg UL for supplemental magnesium; modeled estimate, authors say not enough for guidelines.
Br J Nutr, 2022 · checked 2026-10-07 · we read the abstract
ev-mgbs-08 · Meta-analysis or systematic review · meta-analysis of randomized controlled trials · n = 266
Overall, compared with control intervention for gestational diabetes, magnesium supplementation was able to significantly decrease FPG (MD = -7.33 mg/dL; 95 % CI = -7.64 to -7.02 mg/dL; P < 0.00001) and HOMA-IR (MD = -0.99; 95 % CI = -1.76 to -0.22; P = 0.01), but resulted in no obvious impact on serum insulin (MD = -4.17 μIU/mL; 95 % CI = -8.49 to 0.14 μIU/mL; P = 0.06), preterm delivery (OR = 0.42; 95 % CI = 0.06 to 2.95; P = 0.38), macrosomia (OR = 0.34; 95 % CI = 0.08 to 1.35; P = 0.13) or BMI change (MD = -0.01 kg/m2; 95 % CI = -0.06 to 0.04 kg/m2; P = 0.63).
Who: women with gestational diabetes
Effect: fasting plasma glucose MD -7.33 mg/dL (95% CI -7.64 to -7.02); HOMA-IR MD -0.99 (-1.76 to -0.22); no clear effect on preterm delivery or macrosomia
Certainty: Five small trials, mostly from one country; the very narrow interval suggests a fixed-effect model.
Eur J Obstet Gynecol Reprod Biol, 2024 · checked 2026-10-07 · we read the abstract
ev-mgbs-09 · Randomized controlled trial(s) · randomized double-blind placebo-controlled trial, magnesium oxide 360 mg elemental Mg/day for 4 months · n = 71
Compared to placebo, magnesium supplementation significantly increased plasma magnesium (adjusted mean difference: 0.056 mmol/L, 95% CI: 0.028 to 0.085; p < 0.001) and reduced FPG (adjusted mean difference: -0.497 mmol/L, 95% CI: -0.818 to -0.176; p = 0.003).
Who: community-dwelling older adults in China (mean age 68.7) with prediabetes and plasma magnesium 0.80 mmol/L or less
Effect: fasting plasma glucose adjusted MD -9 mg/dL (95% CI -15 to -3.2; metric: -0.497 mmol/L, -0.818 to -0.176); plasma Mg +1 mg/dL (0.056 mmol/L)
Certainty: Exploratory; baseline imbalance in insulin and HOMA-IR; 65 completed.
Front Nutr, 2026 · checked 2026-10-07 · we read the abstract
ev-mgbs-10 · Randomized controlled trial(s) · randomized double-blind placebo-controlled crossover trial, 15 mmol/day oral magnesium for 6 weeks · n = 14
Despite an albeit modest increase in serum magnesium concentration, oral magnesium supplementation does not improve insulin sensitivity in people with insulin-treated type 2 diabetes and low magnesium levels.
Who: adults with insulin-treated type 2 diabetes and serum magnesium 0.79 mmol/L or less
Effect: clamp M value 4.6 vs 4.4 mg/kg/min (p=0.108); CGM, HbA1c and insulin dose unchanged
Certainty: Very small and short; gold-standard clamp outcome.
Diabetologia, 2024 · checked 2026-10-07 · we read the abstract
ev-mgbs-11 · Observational data · systematic review and meta-regression of prospective cohort studies · n = 637922
After adjusting for age and body mass index, the risk of T2D incidence was reduced by 8%-13% for per 100 mg/day increment in dietary magnesium intake.
Who: prospective cohorts, mainly Asian and US populations
Effect: 8-13% lower risk per 100 mg/day after age and BMI adjustment
Certainty: Observational; magnesium tracks fiber and whole grains, so confounding is likely.
Nutrients, 2016 · checked 2026-10-07 · we read the abstract
ev-mgbs-12 · Position of an expert body · agency fact sheet · n = —
According to the American Diabetes Association, there is currently not enough evidence to recommend using magnesium to improve glycemic control in people with diabetes.
Who: people with diabetes
Effect: agency summary: no recommendation for magnesium for glycemic control
Certainty: ODS restates the ADA position; the cited ADA reference predates the 2022-2026 meta-analyses.
NIH Office of Dietary Supplements, Magnesium fact sheet for health professionals · checked 2026-10-07 · we read the section
ev-mgbs-13 · Position of an expert body · agency fact sheet · n = —
Magnesium deficits and increased urinary magnesium excretion can occur in people with insulin resistance and/or type 2 diabetes
Who: people with insulin resistance and/or type 2 diabetes
Effect: magnesium deficit, higher urinary loss
Certainty: Explains why trials in people with low baseline magnesium may respond differently.
NIH Office of Dietary Supplements, Magnesium fact sheet for health professionals · checked 2026-10-07 · we read the section
ev-mgbs-14 · Position of an expert body · agency fact sheet · n = —
The Tolerable Upper Intake Level for supplemental magnesium is 350 mg for adults, and it ranges from 65 to 350 mg for children and adolescents, depending on age.
Who: adults; children and adolescents
Effect: UL for supplemental magnesium 350 mg/day (adults); food magnesium not counted
Certainty: Many glucose trials used 300 to 600 mg; the dose-response MA estimate at 500 mg is above the UL.
NIH Office of Dietary Supplements, Magnesium fact sheet for health professionals · checked 2026-10-07 · we read the section
ev-mgbs-15 · Position of an expert body · agency fact sheet · n = —
The forms of magnesium that are most commonly reported to cause diarrhea include magnesium carbonate, chloride, gluconate, and oxide
Who: people taking magnesium supplements or medicines
Effect: diarrhea, nausea, cramping at high doses
Certainty: Magnesium oxide was the form in the 2026 prediabetes trial.
NIH Office of Dietary Supplements, Magnesium fact sheet for health professionals · checked 2026-10-07 · we read the section
ev-mgbs-16 · Position of an expert body · agency fact sheet · n = —
The risk of magnesium toxicity increases with impaired renal function or kidney failure because the ability to remove excess magnesium is reduced or lost
Who: people with impaired renal function or kidney failure
Effect: higher risk of magnesium toxicity
Certainty: Diabetic kidney disease is common in long-standing diabetes.
NIH Office of Dietary Supplements, Magnesium fact sheet for health professionals · checked 2026-10-07 · we read the section
ev-mgbs-17 · Position of an expert body · agency fact sheet · n = —
Magnesium may interact with certain medications, such as oral bisphosphonates, tetracyclines, and quinolone antibiotics.
Who: people taking oral bisphosphonates, tetracyclines or quinolones
Effect: drug interactions
Certainty: Doses should be spaced from these drugs.
NIH Office of Dietary Supplements, Magnesium fact sheet for health professionals · checked 2026-10-07 · we read the section

How this page was made. Evidence was extracted from primary sources into cards, every number was re-checked against the source, and the text was written from those cards. Bioma Learn has no human medical reviewer at this time, and we say so rather than invent one. Methodology. This is information, not medical advice.