🧲
Magnesium: the deficiency is real, the promises are not
Research review · PubMed
✍️ Автор: Катя
Magnesium is sold for everything: cramps, stress, insomnia, heart, nerves. We checked each claim against primary sources — the US Institute of Medicine reference intakes, Cochrane reviews, meta-analyses. Some of it held up better than expected. The main reason people buy magnesium in the first place held up worst.
✅
The good news first: the shortfall is genuine
This is the rare case where "everyone is deficient" is not a supplement-seller's invention. The requirement is **400–420 mg** for men and **310–320 mg** for women. Actual intake: men around 340, women around 265. **50–55% of adults** fall below the estimated average requirement. European reference values are set slightly lower — 350 and 300 mg — and deliberately labelled *adequate intake* rather than a full requirement, because the balance data were too thin to support one.
🩸
And the blood test really does miss it
This is confirmed verbatim in the primary source: serum magnesium concentration **"has not been validated as a reliable indicator"** of body magnesium status. The reason is structural: only about **1%** of your magnesium is in blood, while 50–60% sits in bone. The body holds the blood level steady by drawing from bone and cells. Studies have documented cases where cellular magnesium was **abnormally low while serum sat inside the normal range**.
🦵
Now the big one: for leg cramps, magnesium is proven NOT to work
This is the most common reason people buy magnesium, and it produced the worst result. A Cochrane review of 11 trials: the proportion of people who improved was **identical to placebo**. This is **not** "insufficient evidence" — it is high-certainty evidence of **absence** of effect. The authors' wording: it is "unlikely that magnesium supplementation provides clinically meaningful cramp prophylaxis to older adults".
🤰
Pregnancy is the exception, and even there it is shaky
For pregnancy-associated cramps the data could not be pooled at all. One trial found benefit, one did not, a third was inconsistent. And the detail that matters most: **the only trial not rated at high risk of bias trended in favour of placebo**. Certainty of evidence: very low. For exercise cramps and dialysis cramps there is **no evidence at all**.
😴
Sleep: not a sedative, and the evidence base is tiny
The entire evidence base is **three trials in 151 older adults**. Result: falling asleep about **17 minutes faster**, and **total sleep time unchanged**. The authors graded the certainty as low to very low and wrote plainly that "the quality of literature is substandard for physicians to make well-informed recommendations". The newest glycinate trial landed on the edge of significance with a negligible effect size.
❤️
Blood pressure: real, but small
Here the evidence is solid — **34 double-blind trials**. Systolic pressure drops **2 mmHg**, diastolic **1.8**. Statistically dependable. But for scale: the DASH diet delivers 5–11 mmHg. Magnesium is not a replacement for blood pressure medication; it is a small addition. The minimum working dose is around **300 mg daily for a month**.
🍬
Blood sugar: works in people already at risk
In type 2 diabetes magnesium lowers fasting glucose; in at-risk groups it improves the two-hour post-load value. Insulin resistance itself, however, showed only a **non-significant trend**. A healthy person is not going to "improve" anything here.
🚽
The most reliable effect is the one nobody advertises
Magnesium is firmly proven as an **osmotic laxative**. In a constipation trial: **70.6% versus 25%** on placebo. And this is not biology, it is physics — magnesium pulls water into the bowel. Which is exactly why it works most reliably, and exactly why overdosing produces diarrhoea. Worth knowing: **magnesium citrate has no trials in chronic constipation at all** — its evidence base is colonoscopy prep.
⚠️
And a warning that rarely travels outside Japan
In Japan, where magnesium oxide is taken by millions, the regulator has issued **three separate warnings**, and **deaths** from excess blood magnesium are on record. In a hospital study **23%** of patients on magnesium had elevated levels — while it was measured in only **8.9%** of them. Risk factors: **reduced kidney function**, doses above 1650 mg, use beyond a month, older age. With kidney problems, magnesium supplements are not something to take unsupervised.
🤕
Migraine: the one place where form matters
Magnesium officially holds "probably effective" status for migraine prevention; a dedicated systematic review downgraded that to "possibly effective". Working dose: **600 mg of citrate**. The interesting detail is that citrate cut attack frequency by 41.6% against 15.8% on placebo, while **aspartate in a separate trial did nothing**. But the whole base rests on a single 1996 study of 81 people, and no effect on pain severity has been shown.
🧮
How much magnesium is actually in each form
Here is the number that never appears on the label — **percentage of elemental magnesium**: oxide **60%**, carbonate 29%, citrate **16%**, malate 15.5%, bisglycinate 14%, and the fashionable **L-threonate just 8.25%**. So the popular "2000 mg of threonate" is about **150 mg of actual magnesium** — less than one cheap oxide tablet.
🐀
"Threonate for the brain" is still a rat story
The origin of the whole narrative is a study in **rats**. There are four human trials, and they share one feature: every one compared threonate **against a placebo, never against a cheaper form of magnesium**, and every one was **funded by manufacturers**. In the main trial plasma magnesium did not differ between groups. Whether threonate raises magnesium in a human brain is unknown — the authors state that the technology to measure it **does not exist**.
🔄
And the myth in the other direction
"Oxide is only 4% absorbed" is also an exaggeration. That figure comes from a study measuring **magnesium in urine, not absorption**. When it was measured properly, with isotopes, the answer was **22.8%** — on par with expensive diglycinate. Also worth knowing: **no human study has ever compared glycinate with citrate**. Across 35 years there are roughly 13 direct comparisons of forms, with a median of **16 participants**. Organic forms do raise magnesium in blood and urine better — but **not inside cells**, which is where it works.
🎃
Food: one clear champion
Per 100 g: **pumpkin seeds 550 mg**, flaxseed 392, brazil nuts 376, chia 335, cashews about 280, almonds 270, **dark chocolate 70–85% at 228**. Fifty grams of pumpkin seeds is close to a woman's daily requirement. Two corrections: the figures for buckwheat and quinoa (231 and 197) are for **dry grain** — after cooking they fall three to four fold. And **the banana is a myth**: 27 mg per 100 g. Bananas are associated with cramps because of **potassium**, not magnesium.
💊
Who is at risk, and how not to overdo it
The upper limit **for supplements** is 350 mg in the US and 250 mg in the EU. It does not apply to food: 600 mg of magnesium from seeds is fine, 600 mg from a tablet is a laxative. At elevated risk: people on **heartburn medication** (the US regulator warned about this back in 2011, with risk after about a year), on **diuretics**, with diabetes, heavy alcohol use, bowel disease, and older adults. One practical interaction: **tetracycline and fluoroquinolone antibiotics** and **bone medications** must be separated from magnesium by at least two hours.
✅
The bottom line
The shortfall is real and the blood test genuinely will not reveal it. But magnesium is firmly proven as a **laxative**, moderately for **blood pressure** (a modest 2 mmHg) and **glucose in diabetes**, weakly for **migraine** (600 mg citrate). For **leg cramps it is proven not to work**. Sleep and anxiety rest on low-quality evidence. The sensible strategy is **pumpkin seeds, flax, nuts and dark chocolate daily**, with supplements reserved for a specific indication and kept within 250–350 mg.
❓ Часті запитання
Will magnesium help my leg cramps?
No. A Cochrane review found the proportion of people who improved was identical to placebo, at high certainty. That is evidence of absence of effect, not a lack of studies.
How do I find out whether I am deficient?
A routine blood test will not show it: only about 1% of body magnesium is in blood, and the body holds that level steady by drawing on bone. The test is for diagnosing clinical hypomagnesaemia, not for checking your intake.
Which form of magnesium is best?
There are barely any human head-to-head comparisons — around 13 in 35 years, with a median of 16 participants. Organic forms raise blood magnesium better but not the level inside cells. What matters practically is elemental content: 60% in oxide, 16% in citrate, just 8% in threonate.
Is magnesium threonate really better for the brain?
There is no evidence for it. The origin is a rat study, and all four human trials compared threonate with placebo rather than with cheaper forms — and all were manufacturer-funded.
How much magnesium can I take?
The upper limit for supplements is 350 mg daily in the US and 250 mg in the EU. It does not apply to magnesium from food. With reduced kidney function, supplements should only be taken under medical supervision.
Are bananas a good magnesium source?
No — 27 mg per 100 g. Bananas are linked to cramps because of potassium. The real leaders are pumpkin seeds at 550 mg, flaxseed at 392 and nuts.
📚 Дослідження, згадані в статті (14)
- KEY: magnesium reference intakes and why serum is not a marker — US Institute of Medicine (1997) — джерело →
- Magnesium for cramps: 11 RCTs, no effect (Cochrane, 2020) — джерело →
- Magnesium and blood pressure: 34 double-blind RCTs (Hypertension, 2016) — джерело →
- Magnesium and glycaemia: 18 RCTs (European Journal of Clinical Nutrition, 2016) — джерело →
- Magnesium and sleep: 3 RCTs, 151 people, low certainty (BMC Complement Med Ther, 2021) — джерело →
- Magnesium oxide for constipation: 70.6% vs 25% (J Neurogastroenterol Motil, 2019) — джерело →
- Magnesium for migraine: systematic review downgrades to «possibly effective» (Headache, 2018) — джерело →
- Magnesium citrate 600 mg: −41.6% attacks vs −15.8% (Cephalalgia, 1996) — джерело →
- Bioavailability of magnesium forms: direct comparisons are scarce (Curr Nutr Food Sci, 2017) — джерело →
- Isotope measurement: oxide 22.8%, diglycinate 23.5% — no difference (JPEN, 1994) — джерело →
- Magnesium threonate: human trial, plasma magnesium unchanged (J Alzheimers Dis, 2016) — джерело →
- Population magnesium shortfall: NHANES data (Advances in Nutrition, 2020) — джерело →
- Latent deficiency: normal blood levels with depleted tissue (Nutrients, 2018) — джерело →
- Magnesium and anxiety: «the quality of existing evidence is poor» (Nutrients, 2017) — джерело →
Це не медична порада. Маєш діагноз чи обмеження — звернися до лікаря.
Чи була стаття корисною?