Do Multivitamins Work? 84 Trials and 739,803 People Later, the Official Verdict Is “Cannot Be Determined”

Thirty-one percent of American adults take a multivitamin. [2] The body that reviews this evidence for the United States looked at 84 randomised trials covering 739,803 people and concluded that the balance of benefits and harms “cannot be determined”. [1][2]

Not “it does nothing”. Cannot be determined. After 739,803 people. That is a stranger result than a flat debunk and it is worth understanding, because two things in that pile did get a verdict — and one of them is a warning.

The whole shelf, ingredient by ingredient

The good news is real, and much smaller than it sounds

Multivitamin takers had 7% lower odds of being diagnosed with cancer across four trials and 48,859 people. [1] That is a genuine, statistically significant finding and we are not going to wave it away.

Now here is the same finding said the other way. The absolute risk difference — the actual gap in how many people got cancer — was 0.2% to 1.2%. [1]

Between two and twelve people in a thousand. That is what “7% lower” means once you attach it to real people. Both numbers describe the same result; one of them is what a label would print.

And the reviewers add a sentence of their own about it: the evidence for multivitamins “had important limitations”. [1] Which is why the task force, holding a significant result in its hand, still declined to call it. [2]

The memory finding, and what a 0.07 actually is

A large trial called COSMOS tested a daily multivitamin in adults over 60 and looked at thinking and memory. Across 5,203 participants it found a real effect: +0.07 on global cognition and +0.06 on episodic memory, both statistically significant. [3]

Those are effect sizes, and the scale researchers use runs 0.2 for small, 0.5 moderate, 0.8 large. 0.07 is a third of the way to “small”.

The trial itself translated that as being “equivalent to reducing cognitive aging by 2 y”. [3] That sentence is where every “multivitamins turn back your brain clock” headline came from. It is the researchers’ own framing, it is not dishonest, and it travels a great deal further than 0.07 does.

One more thing about that result, because it is described in a way that misleads by accident. It is reported as a meta-analysis of three studies. All three are substudies of the same parent trial, with the same supplement and the same population. That is not three independent replications — it is one trial examined three ways.

The ingredient they tell you to stop

Beta carotene raised the odds of lung cancer by 20% and of dying of a cardiovascular cause by 10%, across trials of 94,830 and 94,506 people. [1]

The task force does not hedge on this one. It recommends against taking beta carotene to prevent heart disease or cancer, concluding that the harms outweigh the benefits. [2] It also concluded there is no net benefit from vitamin E. [2]

Those are the two clear verdicts in the whole review, and they both point the same way. The reviewers also flagged that some supplements were linked to serious harms — hip fracture with vitamin A, hemorrhagic stroke with vitamin E, kidney stones with vitamin C and calcium. [1] “It cannot hurt” is doing a lot of unearned work in this aisle.

And the two biggest sellers did nothing at all

Vitamin D across 27 trials and 117,082 people: no effect on dying of anything. Vitamin E across 9 trials and 107,772 people: the same nothing. [1] Not “too little data”. Measured, at enormous scale, and flat.

Which fits what we found looking at vitamin D on its own and at fish oil, where the real benefit turned out to need a prescription and a blood test.

So should you take one

The honest answer is that nobody can tell you, and that is the finding rather than a dodge. There is a small cancer signal the reviewers do not fully trust, a small cognitive signal from one trial in over-60s, nothing from the two most-sold single vitamins, and one ingredient to actively avoid.

There are two kinds of no evidence, and this is mostly the second kind: not a claim that fell over, but a question that 739,803 people were not enough to settle — because the trials that got run were not the trials that would have settled it.

What follows practically: if you take a multivitamin, the evidence does not tell you to stop, and it does not promise you anything much either. Check that it does not contain beta carotene, which is the one ingredient here with a recommendation against it. And if you are taking it to cover a specific deficiency a blood test found, none of this applies to you — that is treatment, and it is a different question with a different answer.

Where it is worth aiming your annoyance: a category this large, this old and this profitable should have produced a clear answer by now. It has not, and the reason is that nobody selling it needs one. The same pattern shows up on the greens shelf and magnesium is the rare case where something specific survived. Here is how we grade all of it, and here is the shelf ranked by what held up.