A 5-minute read · April 2026

Is there gluten in your prescription?

Most prescription drugs in your medicine cabinet are gluten-free. The catch is that nobody is required to tell you that, and a small fraction of formulations genuinely do contain gluten. Here's the picture in 2026, in plain English.

The regulations: a tale of two coastlines

United States — voluntary, eight years and counting

Gluten in food has been regulated since 2014 — anything labeled "gluten-free" must contain less than 20 ppm. Drugs were specifically excluded from that rule (the Food Allergen Labeling and Consumer Protection Act of 2004 didn't cover medications). The FDA finally issued draft guidance on labeling gluten in oral drugs in December 2017 — Docket FDA-2017-D-6352. As of April 2026, eight years later, it's still draft and still voluntary. Manufacturers may add the statement "Contains no ingredient made from a gluten-containing grain (wheat, barley, or rye)" if it's truthful, but they don't have to. Bills to require disclosure (Rep. Tim Ryan in 2013, 2015, 2019; Sen. Blumenthal S.3021 in 2019) all died in committee.

European Union — mandatory, but only for some drugs

EMA has the opposite stance. When wheat starch is used as an excipient, manufacturers must declare residual gluten content in micrograms per dose, and may only call the product "gluten-free" if it's below 20 ppm. The rule comes from CHMP/704219/2013 and the EU Annex to the 2017 excipient labelling guideline. This is mandatory for centrally-authorised medicines.

The practical gap: EMA's mandatory disclosure only covers ~1,850 centrally-authorised products. Many mainstream prescriptions like atorvastatin, lisinopril, and metformin are nationally authorised through agencies like BfArM (Germany), MHRA (UK), or AIFA (Italy) and aren't in the EMA central database. So the strictest rules apply to the smallest slice of the market.

How many drugs are actually a risk?

Probably less than you'd guess.

The empirical studies broadly agree:

• King 2013, Hospital Pharmacy 48(9):736-43 (PubMed 24421547) surveyed manufacturers of the top 200 prescription drugs by volume. The vast majority confirmed gluten-free; the rest declined to confirm or didn't respond.
• Beyond Celiac 2014 study (FDA-funded, 5,623 survey respondents) — 1,399 (25%) reported a suspected adverse reaction from gluten in medication. Of 39 drugs tested in the lab, most were below the limit of quantification.
• 2024 pediatric study (PMC 12172667) analyzed 689 NDCs across the top 100 pediatric medications: 50.2% confirmed gluten-free, 22.6% gluten-free with no cross-contamination guarantee, 25.7% couldn't confirm, and 1.5% (10 NDCs) actually contained gluten.
• Lizano-Díez 2024 Portugal study (PMC 11958770) found gluten-derived excipients in 44.4% of paracetamol analgesics, 8.2% of ibuprofen NSAIDs, and 0% of antiasthmatics. Most of the gluten was concentrated in liquid pediatric formulations.
• Our own EMA dataset — across all 1,360 parsed EU centrally-authorised leaflets, zero declare wheat starch as an excipient. EU centrally-authorised products skew toward biologics and specialised modern drugs that don't use wheat starch.

A defensible bottom-line estimate is: roughly 1–2% of common adult oral drugs in the US contain any gluten at all, and when present it's usually well below the 10 mg/day threshold that's generally considered safe for celiac patients. (The Catassi 2007 RCT showed villous worsening at 50 mg/day for 3 months; consensus is to keep daily intake under 10 mg.)

Where the risk actually lives

Three real concerns survive that statistical comfort:

1. Liquid and oral suspension formulations, especially pediatric. The Portugal study found 40–50% of liquid paracetamol products had gluten-derived excipients — mostly wheat starch as a thickener. If you give your kid a teaspoon of antibiotic syrup twice a day for two weeks, the cumulative dose can matter even if each milligram is small.
2. Generic substitution. Generics aren't required to use the same inactive ingredients as the brand. The brand may be safe and the generic may not be. The pharmacy may also switch generic manufacturers between fills without telling you.
3. Formulation changes. A manufacturer can change their carrier or excipient supplier without notice. A drug that tested gluten-free last year might use a wheat-derived starch this year. Manufacturer attestations are point-in-time, not warranties.

The practical advice

For most adult oral pills and capsules: low risk, low vigilance needed. For liquids, suspensions, chewables, and especially pediatric formulations: read the inactive-ingredients list, look for "starch" without a source qualifier, and call the manufacturer if you can't confirm. If you switch from brand to generic — or from one generic manufacturer to another — screen the new label even if the active ingredient is identical.

Sources FDA "Gluten in Drug Products and Associated Labeling Recommendations" draft guidance, December 2017; FDA consumer page "Medications and Gluten"; EMA CHMP/704219/2013; King 2013, Hospital Pharmacy 48(9):736-43; Beyond Celiac 2014 FDA-funded study; Spector Cohen et al., Rambam Maimonides Medical Journal 2019; Catassi et al., Am J Clin Nutr 2007; Lizano-Díez et al., 2024; Mangione et al., 2024 (top 100 pediatric medications NDC analysis); Plogsted, glutenfreedrugs.com; empirical analysis of 1,360 EMA centrally-authorised products in this app's own dataset.

How this site is supported

Gluten Screen is free, sign-up-free, and ad-free, and it stays that way deliberately — medical information for people in vulnerable situations should not be gated behind email harvesting or surveillance advertising. The site runs on a small VPS that costs around $15/month, plus a modest amount of time keeping the screening rules up to date. If this tool helped you and you'd like to chip in, you can buy me a coffee ☕. No pressure if not — the tool stays free either way.

Who makes this

Gluten Screen is built and maintained by Ned Walsch, an independent developer. Ned writes the excipient classifier, maintains the 146-entry gluten-flag dictionary, and runs the data pipelines that pull from DailyMed and the European Medicines Agency. The site started because answering “is this prescription safe for someone with celiac?” meant calling manufacturers one at a time, and that seemed like a problem a computer should be able to take off people’s hands.

Ned is not a pharmacist, physician, or dietitian, and Gluten Screen does not give medical advice. What this site does is narrower and, hopefully, more honest than that: it reads the inactive-ingredient lists that manufacturers file with regulators, applies a published set of classification rules to them, and shows you the result along with the reasoning. Every verdict links back to the underlying label so you can check the work. Where a label is ambiguous, the site says so rather than guessing — that is what the “worth a call” category is for.

For any decision that actually matters, confirm with your pharmacist or prescriber. They can see your full medication list and your clinical picture; this tool can only see what is printed on a label.

Editorial policy

How screening decisions get made, in full, so you can judge whether to trust them:

• Primary sources only. Verdicts are derived from manufacturer-filed regulatory documents — FDA Structured Product Labels via DailyMed, and EMA product information leaflets. The site does not source ingredient claims from blogs, forums, retail listings, or other aggregators.
• Published rules, not a black box. The classifier sorts ingredients into contains gluten, worth a call, and nothing flagged using a dictionary that is described on this page. The same rules run for every drug; there are no per-drug manual overrides that would let an unstated judgement slip in.
• Ambiguity is reported as ambiguity. Where a label lists a starch or dextrin without disclosing its botanical source, the result is “worth a call” — never a clean pass. Roughly speaking, the site would rather send you to make an unnecessary phone call than let a wheat-derived excipient through silently.
• No commercial influence. No advertising, no affiliate links, no sponsored placements, no manufacturer funding, and no paid inclusion. No manufacturer has ever been given the ability to change, suppress, or preview a verdict about its own product. The only funding is voluntary reader tips.
• No data collection as a condition of use. The tool requires no account and no email address. Saved medications stay in your own browser.
• Stated limits. A “nothing flagged” result is not a gluten-free certification. Labels do not have to disclose every excipient source, and they say nothing about cross-contamination during manufacturing. Formulations also change without notice, so any result is accurate as of the label revision it was read from.
• Clinical review status. The screening logic has not yet been reviewed by a licensed pharmacist. When that review happens, the reviewer will be named on this page with their credentials and the review date, and the byline on every page will change to reflect it. Until then the site says plainly that it is not clinically reviewed rather than implying otherwise.

Corrections

This site will sometimes be wrong. Labels get misparsed, manufacturers reformulate, and classification rules have edge cases. When that happens, the correction is the priority.

• Reporting an error. Every screening result has a “Tell us what you found” control that submits a correction without requiring any personal information. Reports that a manufacturer has directly confirmed or denied gluten content are the most useful kind, because they carry information no label contains.
• What happens next. Reported errors are checked against the current regulatory label. If the label was misread, the classifier rule is fixed and the fix applies to every drug that shares it, not just the one reported. If the label itself is ambiguous, the entry moves to “worth a call.”
• Substantive corrections are disclosed. If a fix changes a verdict from “nothing flagged” to a flagged state — the direction that could have affected a real decision — it is noted on the affected drug page with the date, rather than being silently edited.
• Source data lag. DailyMed is re-fetched on a twelve-hour cache cycle. EMA, Plogsted, and the Mangione dataset refresh on longer cycles tied to their own publication schedules. Where a dataset is a point-in-time snapshot, the snapshot date is shown next to the result.