Gluten-free vs Mediterranean

What the research found.

• A healthy overall way of eating went with less new rheumatoid arthritis, at an odds ratio of 0.54. It became 0.84 once only the three lowest-risk-of-bias studies were counted. Mediterranean eating on its own came out at 0.88.

  Joerns and colleagues, Seminars in Arthritis and Rheumatism, 2025

• In that review, a healthy eating index approach came out at an odds ratio of 0.60, with the true value running from 0.25 to 1.47. That range crosses one, so it doesn't reach significance. The direction holds up and the size doesn't.

  Joerns and colleagues, Seminars in Arthritis and Rheumatism, 2025

• One trial put 46 women with fibromyalgia on a diet without gluten, dairy, added sugar, or ultraprocessed foods. Their fibromyalgia impact score improved by 19.9 points, against 2.2 on general healthy eating advice. The P value between the groups was 0.001.

  Silva and colleagues, Frontiers in Nutrition, 2022

• In that trial, two measured inflammation markers didn't move at all, because hs-CRP rose slightly in both groups and ESR was unchanged. Every outcome that improved was self-reported. That's the half of the paper nobody quotes.

  Silva and colleagues, Frontiers in Nutrition, 2022

The short answer.

Mediterranean eating is the one with evidence in the diseases people ask about. One review pooled twelve studies and put a healthy overall way of eating at an odds ratio of 0.54 for developing rheumatoid arthritis, which fell to 0.84 in the three lowest-risk-of-bias studies. The Mediterranean approach on its own came out at 0.88.

Going gluten-free has no controlled trial in our source set, because nothing there tested gluten alone in autoimmune disease. The nearest trial took gluten out together with dairy, added sugar, and ultraprocessed foods, and it ran in fibromyalgia rather than an autoimmune condition. Taking just two things off that list could have produced the identical result.

Celiac disease is the exception, and it's a large one. There a gluten-free diet is the treatment rather than an experiment, and one practical point counts most of all. Get tested before you stop eating gluten, rather than after.

Side by side.

What each of these has been tested on

Two eating approaches get recommended for autoimmune disease more than any others, and the evidence behind them isn't comparable. Mediterranean eating has been studied in inflammatory arthritis, with twelve studies pooled. Going gluten-free has been studied in celiac disease, where it's the treatment, and outside celiac disease the picture is much thinner than the confidence around it suggests.

One review looked at a healthy overall way of eating and at the risk of developing rheumatoid arthritis, where the odds ratio was 0.54. That fell to 0.84 once the analysis was cut to the three studies at lowest risk of bias, and Mediterranean eating on its own came out at 0.88. So that's a modest link, from studies that watched people rather than from trials, and it's still more than the other option has.

For gluten on its own in autoimmune disease, our source set holds nothing at all. That's a claim about this library rather than about the whole of the literature, and it's why this page reads the way it does. The nearest trial took gluten out along with three other things, which raises a question that trial can't answer.

The trial everyone quotes, and what it removed

One trial in fibromyalgia put 46 women on one of two diets at random, where one had no gluten, dairy, added sugar, or ultraprocessed foods and the other was general healthy eating advice. The fibromyalgia impact score improved by 19.9 points on the first and by 2.2 on the second, at a P value between the groups of 0.001. Pain, tiredness, sleep, and quality of life all moved in the same direction.

The difficulty is in the design rather than in the result. Four things came out of the diet at once, so the trial shows the set of four helped and says nothing about which one did it. Taking out ultraprocessed foods and added sugar is a big change on its own, either one could account for the whole effect, and gluten need not have contributed anything at all.

There's a second finding in that paper, and it rarely gets quoted. Neither measured inflammation marker moved in either group, meaning neither hs-CRP nor ESR, and every outcome that improved was self-reported. The diet was strict enough that nobody could be kept in the dark about which arm they were in, and the trial ran in fibromyalgia rather than in an autoimmune disease.

Celiac disease is a different question entirely

None of the caution above applies to celiac disease, where the immune answer to gluten is established and a gluten-free diet is the treatment rather than an experiment. So anyone with celiac disease has a clear reason to remove gluten for good and completely. The evidence there isn't in dispute, and this page isn't about it.

What counts in practice is the order things happen in. Celiac testing works by picking up your immune system's answer to gluten you're eating now, so both the antibody blood test and the gut biopsy lose accuracy once gluten has been out for a while. Somebody who cuts gluten first and asks for testing later often gets told to eat it again for several weeks before an answer is possible.

That makes the order the single most useful thing on this page. If celiac disease is possible at all, get tested before making the change. Doing it the other way round turns a simple diagnosis into a much harder one, and it costs weeks of eating the thing you were trying to avoid.

How to run the experiment if you're going to

If you cut a food out for good after a good couple of weeks, that decision rests on the noisiest evidence there is, because inflammatory disease goes up and down on its own. What turns a change into information is a time limit and a plan to put the food back. So take gluten out for a set period, keep a note of what you're tracking, then put it back and see what happens.

A Mediterranean base with a short exclusion on top is a reasonable plan, and it's roughly the form the fibromyalgia trial used. It has one more advantage, and you get it if the exclusion turns out to do nothing. What you're left with is an eating style with evidence behind it, covering your heart and probably your joints, which is a much better place to land than a lasting restriction with no benefit.

Two cautions belong with that. Cutting foods out is the mechanism by which disordered eating gets worse, so anyone with that history should have supervision rather than a self-run experiment. Gluten-free forms of ordinary foods also tend to cost more and hold less fiber, which is a real trade rather than a trivial one.

When each one fits.

• Get tested for celiac disease before cutting gluten, because both the antibody test and the biopsy lose accuracy once gluten is already out of the diet.
• Start Mediterranean if you want the option with evidence behind it, since heart risk is raised in nearly every rheumatic disease and this is the eating style with the most evidence there.
• Consider a supervised gluten trial if you have marked gut symptoms alongside your rheumatic disease, because that group is the closest match to the fibromyalgia trial.
• Treat any exclusion as a time-limited experiment with a planned return of the food, because a diet kept for good after two good weeks rests on the noisiest data there is.
• Don't run an exclusion diet alone if you've had disordered eating, because cutting foods out is the route by which that gets worse.
• Keep taking what your rheumatologist prescribed through either one, since no trial here compared a diet against a disease-modifying drug.

Questions patients ask.