Evidence grades used here: strong trial evidence, mixed, mechanistic and untested, or none.

Two very different reasons to take a nutrient

Replacing a shortfall. You're low, measurably. There's a reason: diet, absorption, a medication, blood loss. You take enough to correct it, recheck, and then stop or drop to a maintenance amount. There's an end point.

Using a nutrient as a drug. You take many times what any diet provides, for an effect that has nothing to do with correcting a deficiency, for as long as you want the effect. That's a legitimate thing to do. But it's pharmacology, and it deserves the same questions as any other medicine: why, how much, for how long, what are the side effects, and how will we know it's working?

The problem isn't either one. It's not knowing which one you're doing.

Two magnesium trials that show the difference

Most supplement trials never check whether anyone was low to begin with. Two magnesium trials happen to sit on either side of the line.

A trial designed as a top-up. A sleep trial in older adults with insomnia only enrolled people whose magnesium intake from food was under three-quarters of the recommended amount (and whose blood magnesium was under 0.95 mmol/L, which is most people). Magnesium (as oxide, 500 mg a day) improved sleep efficiency and how quickly people fell asleep, though not total sleep time (Abbasi et al., 2012). Whatever it showed, it showed it in people who were short to start with. It says little about anyone who wasn't.

A trial that looked like a drug. In an open-label trial of magnesium chloride for mild-to-moderate depression, PHQ-9 scores improved by about 6 points over six weeks. The authors reported similar effects whatever people's starting magnesium level (Tarleton et al., 2017). If the benefit doesn't depend on being low, it isn't correcting a shortfall. There was no placebo, though, so part of that effect may be expectation. Grade: mixed.

Case 1: The migraine supplement

A popular migraine supplement, sold as Dolovent in the UK and Migravent in Germany, gives 600 mg of magnesium, 400 mg of riboflavin (vitamin B2) and 150 mg of coenzyme Q10 a day, plus a low-dose multivitamin. Its website says it can be used long-term, and that normal levels "may be restored in as little as a week". It gives no source for that.

What its own trial found. 130 adults with migraine took it or a placebo for three months. Migraine days fell from 6.2 to 4.4 a month on the supplement, and from 6.2 to 5.2 on placebo. That difference, the trial's main outcome, was not statistically significant. Pain intensity and a headache-burden score did improve (Gaul et al., 2015). The authors put the miss down to too few people. That may be right, but a trial that misses its main outcome hasn't shown it.

Three more details from the full paper are worth knowing. The trial was run with the manufacturer's involvement: the paper describes compliance being monitored by "delegates of the sponsor", and one of its three authors, Ulrich Danesch, is employed by the sponsor, Weber & Weber, and coordinated the trial's monitoring and statistical analysis. About one in eight people on the supplement noticed bright yellow urine from the riboflavin, which can tell someone they're not on placebo. And 17 of the 130 were left out of the efficacy analysis for protocol violations. Grade: mixed. People may genuinely feel better on it; the trial didn't show fewer attacks.

The magnesium. The classic magnesium migraine trial gave 600 mg a day of trimagnesium dicitrate for 12 weeks. Attacks fell by 41.6% against 15.8% on placebo, and 18.6% of people had diarrhoea (Peikert et al., 1996). A second placebo-controlled trial, the same year, found no benefit and was stopped early (Pfaffenrath et al., 1996). Grade: mixed.

For what these trials mean if your migraine shows up as dizziness rather than headache, see Dizzy, Foggy, No Headache.

Neither trial measured magnesium status before or after. So we don't know whether the people who improved were low to begin with, or whether magnesium at that dose was acting more like a drug.

There's a second catch. Serum magnesium, the usual blood test, holds less than 1% of the body's magnesium, and the body works hard to keep it in range. People with depleted tissue stores can have a normal blood result. A 2026 review argues that this is why supplement trials so often report clinical benefit without any change in the blood test (Chandarana & Butani, 2026; the authors run a diagnostic centre and a biotech company). So "were you low?" is harder to answer than it sounds. That's a reason to be careful, not a reason to skip the question.

Is 600 mg a lot? The EU's Scientific Committee on Food set the upper level for magnesium from supplements, on top of food, at 250 mg a day for adults, because more can cause diarrhoea (BfR). The UK's Expert Group on Vitamins and Minerals set a guidance level of 400 mg. The effect is reversible within a day or two in healthy people with normal kidneys. So 600 mg is above both, and the trials used it on purpose. (My own magnesium page keeps to 250 mg by default; it was corrected in October.) That's using magnesium as a treatment, and it's why kidney function matters before long-term use: the kidneys clear the excess.

The riboflavin. One small trial (55 people) found 400 mg of riboflavin a day reduced migraine: 59% of people halved their headache days, against 15% on placebo (Schoenen et al., 1998). Grade: mixed — one small positive trial, not yet strong. The absorption study is humbling, though. In healthy adults, the most riboflavin absorbed from a single dose was about 27 mg (Zempleni et al., 1996). Most of a 400 mg dose passes straight through, which is why urine turns bright yellow. The trial still found a benefit, so absorption limits don't disprove it. But they do mean nobody can tell you exactly how it works. The mechanism: mechanistic and untested.

Case 2: The doctor and his vitamin B3

The dose in the trials is 500 mg of nicotinamide, twice a day. Nicotinamide is a form of vitamin B3 that, unlike nicotinic acid, doesn't cause flushing.

The trial. 386 people who had already had at least two non-melanoma skin cancers in the previous five years took nicotinamide or a placebo for a year. New skin cancers were 23% lower on nicotinamide. And the line that matters most: "there was no evidence of benefit after nicotinamide was discontinued" (Chen et al., 2015). Grade: strong trial evidence, for that group.

The trial that didn't work. In organ-transplant recipients, whose immune systems are deliberately suppressed, the same dose made no difference: 207 new skin cancers on nicotinamide against 210 on placebo. That trial stopped early because it couldn't recruit enough people (Allen et al., 2023).

The large record review. Among more than 33,000 US veterans, nicotinamide users had 14% fewer subsequent skin cancers, and 54% fewer when they started after their first one (Breglio et al., 2025). Grade: observational, supportive. People who choose to take a supplement differ in other ways, however carefully they're matched.

Is 1,000 mg a lot? Yes, by any upper level. The EU's Scientific Committee on Food set the upper level for nicotinamide at about 900 mg a day for adults, and the UK's Expert Group on Vitamins and Minerals set a guidance level of about 500 mg a day from supplements (both as quoted in the Norwegian food-safety committee's 2017 assessment). The trial dose sits above both. That doesn't make it dangerous at that dose for a year under a dermatologist. It does make it a drug dose, not a top-up.

What that means. Nicotinamide at 1,000 mg a day isn't correcting a B3 deficiency. Nobody in these trials was selected because they were deficient, and the average UK adult already gets more than twice the recommended amount of B3 from food (Derbyshire, 2018, from the National Diet and Nutrition Survey; the analysis was funded by the Health & Food Supplements Information Service). It's working like a drug: in a defined high-risk group, only while you take it.

So for someone who's had skin cancers, it's a reasonable, evidence-based choice to discuss with a dermatologist. For someone who hasn't, there's no trial either way. Grade for people without previous skin cancers: none. That's not proof of harm; it's an absence of evidence of benefit.

The questions to ask about any long-term supplement

  1. Is this replacing a shortfall, or is it a treatment? If it's replacing a shortfall, what showed you were low?
  2. What dose did the trials use, and in what form? Magnesium glycinate isn't magnesium dicitrate. Two drops isn't 400 mg.
  3. Who was in the trial? People with frequent migraine? People with previous skin cancers? People who were low to start with? Are you like them?
  4. How long, and what happens when you stop? If benefit disappears on stopping, you're taking a drug. Fine, but know it.
  5. What would tell you it isn't working? Write it down before you start: attacks per month, new lesions, a blood result. Then check.
  6. What does it cost you? Money, side effects, interactions, kidney load, and the attention it takes away from whatever caused the problem in the first place.

That last one is the one I care most about. A supplement that settles a symptom can also stop anyone asking why the symptom was there. Why More of a Thing Usually Does Nothing covers the other side of this: for most nutrients, once you're replete, more adds nothing.

Where testing fits

Testing can't answer every question here; serum magnesium is a good example of its limits. But it separates the two situations better than guessing does:

Declared interest: testing is what I sell. No trial has shown that testing before supplementing improves migraine or skin-cancer outcomes. What it does is stop you treating a deficiency nobody measured.

If you get a new kind of headache after 50, a sudden severe headache, or headache with weakness, confusion or loss of vision, see a doctor urgently or call 999.

Your next step

Free: Health Pattern Finder: about 5 minutes, to see which body systems to look at first.

Test / talk: Blood testing options, to find out whether you're actually short of something; or a free 20-minute call to talk through a supplement you're already taking.

I sell the testing above. The article's grades apply to it too.

References

  1. Abbasi B, et al. The effect of magnesium supplementation on primary insomnia in elderly: a double-blind placebo-controlled clinical trial. J Res Med Sci 2012;17(12):1161–1169. PMID 23853635. PMC3703169
  2. Allen NC, et al. Nicotinamide for skin-cancer chemoprevention in transplant recipients. N Engl J Med 2023. PMID 36856616. doi:10.1056/NEJMoa2203086
  3. BfR. Maximum daily level for magnesium in food supplements (SCF upper level, 250 mg/day supplemental). bfr.bund.de. Accessed 30 Sep 2026.
  4. Breglio KF, et al. Nicotinamide for skin cancer chemoprevention. JAMA Dermatol 2025;161(11):1140–1147. PMID 40960808. doi:10.1001/jamadermatol.2025.3238
  5. Chandarana J, Butani K. Why serum magnesium fails: a narrative review. J Am Nutr Assoc 2026. PMID 42647115. doi:10.1080/27697061.2026.2719468
  6. Chen AC, et al. A phase 3 randomized trial of nicotinamide for skin-cancer chemoprevention. N Engl J Med 2015. PMID 26488693. doi:10.1056/NEJMoa1506197
  7. Derbyshire E. Micronutrient intakes of British adults across mid-life: a secondary analysis of the UK National Diet and Nutrition Survey. Front Nutr 2018;5:55. PMID 30073167. doi:10.3389/fnut.2018.00055. Funded by the Health & Food Supplements Information Service.
  8. Gaul C, Diener HC, Danesch U. Improvement of migraine symptoms with a proprietary supplement containing riboflavin, magnesium and Q10. J Headache Pain 2015;16:516. PMID 25916335. doi:10.1186/s10194-015-0516-6
  9. Peikert A, et al. Prophylaxis of migraine with oral magnesium. Cephalalgia 1996;16(4):257–263. PMID 8792038. doi:10.1046/j.1468-2982.1996.1604257.x
  10. Pfaffenrath V, et al. Magnesium in the prophylaxis of migraine — a double-blind placebo-controlled study. Cephalalgia 1996;16(6):436–440. PMID 8902254. doi:10.1046/j.1468-2982.1996.1606436.x
  11. Schoenen J, et al. Effectiveness of high-dose riboflavin in migraine prophylaxis. Neurology 1998;50(2):466–470. PMID 9484373. doi:10.1212/wnl.50.2.466
  12. Tarleton EK, et al. Role of magnesium supplementation in the treatment of depression: a randomized clinical trial. PLoS One 2017;12(6):e0180067. PMID 28654669. doi:10.1371/journal.pone.0180067
  13. VKM (Norwegian Scientific Committee for Food and Environment). Assessment of intake of nicotinic acid and nicotinamide in relation to tolerable upper intake levels. VKM Report 2017:27. vkm.no PDF. Quotes SCF 2002 (nicotinamide UL 12.5 mg/kg bw/day, ~900 mg/day) and EVM 2003 (guidance ~500 mg/day supplemental).
  14. Zempleni J, et al. Pharmacokinetics of orally and intravenously administered riboflavin in healthy humans. Am J Clin Nutr 1996;63(1):54–66. PMID 8604671. doi:10.1093/ajcn/63.1.54