How Often Does "Gluten-Free" Restaurant Food Contain Gluten?
Researchers at Columbia University pooled 5,624 real-world tests taken by diners in restaurants. Roughly a third of dishes labeled gluten-free came back positive — and pizza and pasta were the worst offenders by a wide margin.
Packaged food is the easy case. A barcode gets you an ingredient list, a manufacturer and a regulated label claim. A plate that arrives at a restaurant table has none of that, which is why eating out is consistently described as the hardest part of a gluten-free diet. Until recently there was very little data on how often the "gluten-free" line on a menu actually holds up.
What the study did
A team at the Celiac Disease Center at Columbia University Medical Center, led by Benjamin A. Lerner with senior author Benjamin Lebwohl, took a different route to the problem. Instead of running a small controlled sampling study, they analyzed data that diners were already generating.
Users of a portable gluten sensor — a consumer device that runs a small sample of food through an antibody test and returns a positive or negative result in a few minutes — could upload each test along with the date, time, food item, and restaurant name and address. Over an 18-month period, 804 users contributed 5,624 tests of restaurant food that had been labeled or described as gluten-free. The results were published in the American Journal of Gastroenterology in May 2019.
What they found
Across everything tested, gluten was detected in about 32% of dishes labeled gluten-free. Breaking it down by food type made the picture sharper:
| Category | Tests positive for gluten |
|---|---|
| All gluten-free labeled items | about 32% |
| Gluten-free pizza | about 53% |
| Gluten-free pasta | about 51% |
| Breakfast items | about 27% |
| Lunch items | about 29% |
| Dinner items | about 34% |
Pizza and pasta stand out, and the reason is not mysterious. Both are typically cooked in equipment that also handles their wheat-based versions: the same oven deck dusted with semolina, the same pot of boiling water, the same colander. Gluten-free pizza dough placed on a surface that just held regular dough picks up flour, and flour is remarkably mobile in a working kitchen.
The meal-time gradient is smaller but consistent — detection rose from breakfast through dinner. A busier service, more dishes in flight and more shared equipment is the intuitive explanation, though the study measured the pattern rather than proving the cause.
The caveats matter
Two limitations are worth stating plainly, because the headline number gets quoted without them.
First, a positive result is not the same as a violation. The device registers gluten at roughly 5 to 10 parts per million, while the regulatory threshold for a gluten-free claim in both the United States and the European Union is below 20 ppm. Some dishes that tested positive would still have been legally and, for many people, practically gluten-free. The researchers noted this themselves.
Second, the sample selected itself. These were tests people chose to run and chose to upload. Someone who suspects a dish is risky is more likely to test it than someone eating at a dedicated gluten-free bakery, and a surprising positive is arguably more likely to get uploaded than an unremarkable negative. The 32% figure describes this dataset, not a random sample of restaurants.
What survives those caveats is still substantial: cross-contact in restaurant kitchens is common rather than exceptional, and it clusters in predictable places. That is a useful thing to know even if the precise percentage is soft.
Why this is a kitchen problem, not a menu problem
Lebwohl's own reading of the results was that the contamination is largely inadvertent — restaurants are not putting wheat in the gluten-free pasta, they are cooking it in water that already has wheat pasta in it. His suggested remedy was better education for the people preparing the food, not tighter menu rules.
That reframes what a diner can usefully ask about. "Is this gluten-free?" invites a yes from someone reading a recipe. "Is it cooked in separate water, or in a separate fryer?" asks about the step where the study suggests things actually go wrong.
The source
Lerner BA, Phan Vo LT, Yates S, Rundle AG, Green PHR, Lebwohl B. Detection of Gluten in Gluten-Free Labeled Restaurant Food: Analysis of Crowd-Sourced Data. American Journal of Gastroenterology, May 2019; 114(5): 792–797.
Read the paper (DOI: 10.14309/ajg.0000000000000202) • PubMed record • Columbia Mailman School summary
Common questions
How often does restaurant food labeled gluten-free actually contain gluten? +
In a Columbia University analysis of 5,624 crowd-sourced tests published in the American Journal of Gastroenterology in 2019, about 32% of restaurant dishes labeled gluten-free returned a positive result on a portable gluten sensor. The figure describes that dataset, which was self-selected, rather than restaurants in general.
Which gluten-free restaurant dishes tested positive most often? +
Pizza and pasta. Gluten was detected in about 53% of gluten-free labeled pizza samples and about 51% of pasta samples — roughly double the overall rate across all foods in the study.
Does a positive gluten test mean the food was unsafe? +
Not necessarily. The device used in the study registers gluten at roughly 5 to 10 parts per million, while the US and EU regulatory threshold for a gluten-free claim is under 20 ppm. A positive result therefore includes some foods that would still meet the legal standard, and the researchers noted this as a limitation.
Did the time of day make a difference? +
Yes, modestly. Gluten was detected in about 27% of breakfast items, 29% of lunch items and 34% of dinner items. One proposed explanation is that dinner service is busier and involves more shared equipment.
Why does cross-contact happen in restaurant kitchens? +
Shared fryers, shared pasta water, shared pizza ovens, shared cutting boards, flour dust in the air and utensils moved between dishes are all common routes. The study authors suggested better education for food preparers as part of the solution.
Check a product in seconds
GlutenScanning reads the barcode, parses the ingredient list and flags gluten plus 60+ other allergens. Free, no account required.
Download on the App Store: GlutenScanning for iPhone