Health Britannica · Research · Dataset · NIH labels read 10 January 2026
| Nutrient | At or over the limit, minimum daily servings | At or over, maximum daily servings |
|---|---|---|
| niacin | 36% (8,870 of 24,456) | 39% (9,504) |
| magnesium | 15% (4,222 of 28,569) | 17% (4,952) |
| vitamin D | 8.6% (2,347 of 27,419) | 9.3% (2,551) |
| vitamin B6 | 6.1% (1,817 of 29,582) | 6.8% (1,997) |
| zinc | 6.0% (1,576 of 26,147) | 7.6% (1,976) |
| manganese | 5.1% (808 of 15,704) | 5.7% (887) |
| iron | 3.8% (501 of 13,237) | 4.2% (553) |
| folic acid | 3.2% (750 of 23,507) | 3.9% (922) |
Compiled and checked by Vincent Wesley Couey (ORCID 0009-0005-6869-308X) from the NIH Dietary Supplement Label Database and the NIH Office of Dietary Supplements fact sheets. Not yet reviewed by a clinician. Built 12 November 2026.
Niacin, magnesium and vitamin D have the largest shares of labels at the adult limit, and the share falls steeply after them. Each bar is the share of assessable labels at or over the limit at the label's minimum daily servings.
385,490 label and nutrient pairs had an assessable amount. Nutrients with fewer than 1,000 assessable labels are in the file but not drawn.
A label was measured by its first listed serving, multiplied by the daily servings it states, then compared with the NIH adult upper limit for ages 19 to 50. Only the nutrient that counts toward the limit was added: preformed vitamin A and not carotene, supplemental vitamin E, folic acid and not food folate, nicotinic acid and nicotinamide for niacin, supplemental magnesium. A label that does not state the form is counted as a range, and the page's counts use the low end. Units convert by the NIH sheets' own rules.
The measurement was checked by a separate program. The method was written and hashed before any label was compared with any limit (sha256 89d8bb80b9a14c1e...). A second implementation, written separately and reading the raw NIH records, recomputed 500 random labels and matched the pipeline on all 461 simple-mass nutrient values, with 0 mismatches. The control did not cover the form-dependent nutrients (vitamin A, D, E, folic acid, niacin), which rest on unit tests built from the NIH sheets' worked examples, and it did not test labels that enter a compound's weight under the nutrient's name.
The most common mistake reads a label at a limit as a product that is unsafe. The limit is set from evidence on harm at the group level.
"A third of niacin supplements are over the safe limit."
Read it as a third of niacin labels state a serving that reaches the NIH upper limit, add what a person eats, and ask a pharmacist or doctor what suits them. See the niacin fact sheet, the vitamin D fact sheet, the magnesium fact sheet, the folate fact sheet and the vitamin A fact sheet.
| File | What is in it | Size |
|---|---|---|
| aggregates.csv | Per nutrient: labels assessed, how many sit under half, half to the limit, may exceed, at or over (at minimum and at maximum daily servings), and why some labels cannot be assessed | 4 KB |
| aggregates-by-entry-year.csv | Labels at or over the limit at minimum servings, by nutrient and the year the label was entered | 6 KB |
| ul-limits.csv | The NIH upper limits used: one row per nutrient, age band and group, with the scope sentence and the sheet date | 97 KB |
| PROTOCOL-DOSE-VS-LIMIT-2026-11-11.md | The method, written and hashed before any label was compared with any limit | 7 KB |
| control-result.json | The independent recomputation of a random sample of labels | 285 B |
| data.json | Every figure on this page, as the generator read it | 5 KB |
| table_labels.csv.gz | All 214,780 labels: name, brand, barcode, entry date, market status (Zenodo only) | 7.7 MB |
| table_ingredients.csv.gz | About 2.0 million ingredient rows: form, amount, unit, percent Daily Value, serving size (Zenodo only) | 41.0 MB |
The full set, with every label table, is on Zenodo as doi:10.5281/zenodo.24148920. Look up one product with the Supplement Stack Checker.
Couey, V. W. (2026). Supplement Label Atlas 2026: how many labels reach an NIH upper limit in one day's servings (Version 1.0) [Data set]. Health Britannica. https://doi.org/10.5281/zenodo.24148920
A label reaches the limit when its first listed serving, multiplied by the daily servings the label states at the minimum, supplies at least the NIH adult upper limit for that nutrient from the supplement alone. Where a label leaves out the form a limit depends on, the count uses the low end of a range, so a label counts as reaching the limit only when even the low end does.
No. The NIH upper limit is a level unlikely to cause harm in nearly everyone, not a threshold at which harm is certain, and some people take higher amounts under medical advice. The Atlas counts labels and makes no finding about any product.
The niacin limit applies to supplemental niacin and is 35 mg for adults, while many B-complex and multivitamin labels state 50 mg or more per serving. The count is of labels, not of people taking them.
Yes. The Supplement Stack Checker searches the same 214,780 labels by name or barcode and adds the amounts across everything you take.
The labels are the NIH export dated 10 January 2026. A label is what the maker declared on the day it was filed, and a product may have changed or left the market since; the tables carry each label's market status.
Yes. The compilation is published under a Creative Commons Attribution licence, and the underlying NIH records and fact sheets are in the public domain.
None yet. A correction is listed here with its date, and earlier versions stay available under the same DOI. Further reading: the National Academies Dietary Reference Intakes; the FDA nutrition labeling rule, 21 CFR 101.9; the FDA dietary supplements page; report a problem through the Safety Reporting Portal.