Diet with Hashimoto's: what the studies show and what is experience

You have received the diagnosis. At home you type two words into the search bar: Hashimoto diet.

And then a list appears. No gluten. No cabbage. No milk. No soy. No iodine. No nightshades. What is left at the end is rice with zucchini, and even about that you are no longer sure.

The next day you stand in the supermarket and do not know what to put in the cart.

Many people know this pattern. In my consultations I hear it almost every week, usually in a sentence that sounds roughly like this: I have already left out so much and it has not got any better.

Hence this text. It hands out no prohibitions. It lays open what has actually been studied in Hashimoto's, with numbers, study sizes and the places where the data contradict one another. In some places the result is more reassuring than you think. In one place it is surprisingly concrete.

Why diet can dock onto an autoimmune condition at all

The first question is rarely the one about broccoli. It is: what does my food have to do with my immune system? The thyroid sits in the neck, the gut lies half a metre lower. Why should one speak with the other?

Because the larger part of your immune system works down there. The gut lining is a paper thin border between your inside and everything you eat. At this border it is decided every day what is harmless and what triggers an alarm. Picture a border station that is permanently open and in which the papers have to be checked very quickly.

A scoping review from 2025 has collected where this border station and the thyroid touch each other.

The term molecular mimicry sounds technical, but it means something very vivid. Two people wear the same coat. The guard recognises the coat, not the face. If a bacterial protein resembles a protein of your own body, a defence that has once started can hit both. Why the immune system enters this state in Hashimoto's at all is a chapter of its own, and I have described it in detail elsewhere: Hashimoto's and the immune system.

The second connection is less spectacular and in practice at least as important. A review from the endocrinology department in Graz describes the thyroid gut axis from a completely different direction [Mechanism Review]. The composition of the gut flora influences how well you absorb micronutrients in the first place. Iodine, iron and copper are needed for hormone synthesis, selenium and zinc for the conversion of T4 into the more active T3, vitamin D for immune regulation. Whoever absorbs poorly has less building material, quite independently of what was on the plate.

That brings us to the basic logic of this article. In Hashimoto's, diet can act at three points: at the barrier and the immune regulation in the gut, at the supply of building blocks, and at the absorption of the medication. The third point is the best documented one, and it comes in section seven.

The mechanics of the gut barrier, meaning zonulin, permeability and how measurable they are, I deliberately do not unfold here. That is covered in detail in Leaky gut and intestinal permeability. The same goes for the question of what gliadin and the so called amylase trypsin inhibitors in wheat might do beyond celiac disease: Gluten and gliadin without celiac disease. And because diet is not the only environmental factor acting on this axis, the chapter Heavy metals and Hashimoto's belongs in the picture too.

Gluten, part one: celiac disease is not a footnote

Almost every discussion about Hashimoto's and gluten begins in the wrong place. It begins with the question of whether wheat harms the immune system. It should begin with a different question: have you actually been tested?

Because there is a group for which gluten is demonstrably a problem. And in Hashimoto's this group is clearly larger than in the general population.

This clustering is no coincidence. Autoimmune conditions appear in groups. A meta-analysis of 180 papers with a combined 293,889 people with type 1 diabetes shows the pattern very clearly: hypothyroidism in 9.8 percent, celiac disease in 4.5 percent, autoimmune gastritis in 4.3 percent, vitiligo in 2.4 percent [Meta-analysis, k=180, n=293,889]. When the immune system has lost tolerance at one place, a look at the other places is worthwhile.

The second half of this section is the more important one, and it is about an order of steps.

The updated guideline of the American College of Gastroenterology from 2023 describes celiac disease explicitly as a multisystem condition and not as an isolated bowel disease [Guideline]. Screening runs through antibodies in serum, above all against tissue transglutaminase. In most adults a small bowel biopsy follows for confirmation. Treatment is a strict gluten free diet with lifelong medical support.

And now the point at which the German guide landscape does real damage: these antibodies disappear on a gluten free diet. They are, after all, a reaction to gluten. Take the gluten away and the reaction falls, and the test becomes unusable.

Gluten, part two: what happens without celiac disease, honestly counted

Now it gets uncomfortable. Not because the answer is bad, but because there is none.

Many people with Hashimoto's report that they feel better without wheat. I take that seriously. It is still something different from proof that the condition itself changes in the process. Let us look at what was actually measured.

The paper most cited in the German language internet comes from Poland and is almost always reproduced incorrectly.

Two meta-analyses have tried to make an overall picture out of the available individual studies. They arrive at different results, and this contradiction is the real information gain of this section.

The most interesting side finding sits in the subgroup analysis by Piticchio. The striking scatter between the studies dissolved as soon as one separated whether a gluten associated condition was involved or not. The authors phrase their own conclusion cautiously: the results point towards a positive effect, above all in people with gluten associated complaints, but the evidence is not sufficient to recommend this approach to everyone with Hashimoto's.

There is exactly one paper that tests gluten free directly against another dietary pattern. It is small, but it asks the question differently.

This is the methodological objection that runs through almost all dietary studies in Hashimoto's and that I will pick up once more in this text: in very many of these papers body weight fell at the same time. Not a single design separates cleanly what is attributable to leaving out a food and what to the reduction in energy.

Whether and how antibodies can be influenced at all in a targeted way, and what a falling titre means in practice, is a question of its own. It is covered in Lowering Hashimoto antibodies. And anyone who wants to know what is discussed about wheat beyond celiac disease will find it in Gluten and gliadin without celiac disease.

The autoimmune protocol in Hashimoto's: what was measured and what was not

At some point in every search the autoimmune protocol appears, usually abbreviated as AIP. And with it a promise that carries a lot of force in forums: people report that with it they felt alive again for the first time in years.

A brief orientation, without unfolding the topic here: AIP is a staged elimination diet from the paleo world. In a first phase grains, legumes, dairy products, eggs, nightshades, nuts, seeds, alcohol and additives are dropped. After that foods are reintroduced step by step and it is observed what is tolerated. The mechanics of this reintroduction, the lectin discussion and the nightshade question I have described fully in Paleo, AIP and autoimmune conditions. Here it is exclusively about what was measured in Hashimoto's.

And those are exactly two studies.

When this study appears in guides, it usually ends at the CRP. The negative finding is missing. Yet that is precisely the actual news.

On top of that come the limits of the design: seventeen people, no control group, no randomization, no blinding, ten weeks. And the intervention was a package of diet, weekly coaching and group belonging. How much of it falls to the diet cannot be determined from this paper. Ten weeks of intensive support do something with people, even without a single food being dropped.

A critical review from Thessaly in 2024 appreciates the approach and at the same time names the calculation that goes with it [Review]. The available clinical studies are small. And the burden of a very restrictive diet has to be weighed against the benefit that can be expected.

This burden is real and is regularly left out in guides. An elimination phase costs time, money and social mobility. It can create nutrient gaps. And in some people it can set off a relationship with food that becomes narrower and more anxious than it was before. If you notice that the lists in your head grow louder than your hunger, that invitations become a problem or that control feels good, then that is a signal, not a success. The text Understanding eating disorders describes how fluid this transition can be.

And now you know why with AIP I do not speak of a therapy but of a time limited attempt with an open outcome.

Iodine in Hashimoto's: why different rules apply here

Iodine is the topic where two camps face each other most irreconcilably. One side says: the thyroid needs iodine, so take iodine. The other says: in Hashimoto's iodine is poison, avoid everything.

Both sentences are too coarse. The data say something more precise, and it is about amounts.

Mechanistically the direction is understandable. A review from Athens describes what could happen in susceptible people under iodine excess [Mechanism Review]. The number of T helper 17 cells in the thyroid rises, the development of regulatory T cells is inhibited. On top comes an abnormal expression of a signalling molecule called TRAIL in the thyroid cells, which promotes programmed cell death. In the mouse model it was also shown that high iodine intake makes the thyroglobulin molecule more immunogenic and lets more reactive oxygen species arise [In vivo, mouse]. The last part comes explicitly from animal experiments and is not documented that way in humans.

Now comes the counter check, and it is rarely told.

Iodine physiology itself, meaning requirement, uptake, measurement and the question of how well supplied Germany is, I do not unfold here. That is covered in Placing iodine in context.

In pregnancy the logic turns around

This paragraph is short and important. The guideline of the American Thyroid Association from 2017 on the thyroid in pregnancy and the postpartum period regulates reference values, iodine supply, handling of antibodies and screening in its own right [Guideline]. In pregnancy different target values for TSH apply. And the iodine question turns: there, iodine deficiency is the dominant risk, because the child depends on the maternal supply.

Concretely that means: if you are pregnant, want to become pregnant or are breastfeeding, no point in this article is a reason for a self experiment. Neither an iodine elimination nor an elimination diet nor any change to the medication. That belongs in medical care built for this situation.

Cabbage, broccoli and soy: the most common question in the consultation

When I ask people with Hashimoto's what they have left out since the diagnosis, very often the same answer comes: cabbage. And then the postscript, half apologetic: I know that sounds odd, but I do not dare any more.

This question is not silly. It has a real core, and that core is over a hundred years old. It has only grown much larger along the way than the data can carry.

The term is goitrogens, meaning goiter forming substances. Cruciferous vegetables such as cabbage, broccoli, cauliflower, Brussels sprouts, kale, radish and mustard contain glucosinolates. When they are chopped, breakdown products arise from them that can hinder the uptake of iodine into the thyroid. So far it holds.

What is decisive is where this observation comes from.

A Polish review from 2024 took on exactly this question, namely whether excluding cruciferous vegetables in Hashimoto's is justified [Review]. The answer of the authors from Białystok: glucosinolates can inhibit iodine uptake, yet the data do not justify a blanket exclusion from the diet in Hashimoto's. The full text is in Polish, which is why I draw on this paper only for orientation and not for quantities.

On cooking, because that is the second most common follow up question: heating lowers the content of glucosinolates and their breakdown products. That is chemically easy to follow. For Hashimoto's this effect is not quantified though. There is simply no study comparing cooked against raw cabbage in autoimmune thyroiditis. Anyone drinking a litre of green smoothie with raw kale every day is in a different order of magnitude from someone eating broccoli twice a week. That is the honest state of things.

Soy, and where the warning actually comes from

With soy the numbers are more precise, and they are astonishingly undramatic.

A systematic review by Messina and Redmond had already screened fourteen intervention studies back in 2006 and recorded something remarkable in the process: in not a single one of these studies was thyroid function the primary endpoint [Systematic Review]. With one exception, either no effects or very small changes were found. The authors' conclusion: in euthyroid people with good iodine supply there is hardly any sign of impairment. For people with hypothyroidism there are indications that soy could raise the hormone requirement through inhibited absorption, but they too would not have to avoid soy. The theoretical residual concern comes from in vitro and animal data [In vitro].

There is exactly one paper that supports a real warning, and it has to be delimited precisely.

The intake gap: the best documented point that almost nobody writes about

For this article I worked through five of the most read German language pages on Hashimoto's and diet. On not a single one did this section appear.

That is remarkable, because among the dietary questions in Hashimoto's this point rests on the most solid randomized foundation, and the effect size there is large enough that it can make a difference in everyday life.

The second building block is an Italian investigation that grew out of a real consultation observation. Eight women had unexpectedly high TSH values, and what they all had in common was that they took their tablet with coffee or espresso.

Coffee is only one entry on a longer list. A review from Los Angeles has collected what can disturb the absorption of thyroxine [Review]. It is the most practical source in this whole article.

This table reproduces exclusively numbers and lists from the cited studies. It is not an instruction for action and not a personal recommendation. What applies to you is decided in medical management.

Two points from this list deserve a sentence of their own.

Iron. Iron sulfate is among the best known disruptors of thyroxine absorption. At the same time iron deficiency is common in people with Hashimoto's, because both autoimmune gastritis and celiac disease can impair iron absorption. So it can be sensible to supplement iron and necessary to separate it in time from the thyroid tablet. How closely iron, thyroid and sleep hang together is covered in Iron deficiency, thyroid and sleep.

Lactose. A Turkish paper examined 83 people with Hashimoto's on levothyroxine and found lactose intolerance in 75.9 percent, an unusually high value for this population [Case Series, n=50]. In those affected, TSH fell after eight weeks of a low lactose diet from 2.06 ± 1.02 to 1.51 ± 1.1 mIU/l, and in the subclinically hypothyroid ones even from 5.45 ± 0.74 to 2.25 ± 1.88. In the people without lactose intolerance nothing changed. The paper was not randomized and the group was small. The finding should be read as “worth checking if the values fluctuate despite correct intake”, not as “everyone should drop milk”.

What diet can do, what it cannot do, and what I observe in the consultation

That leaves the question most people came into this article with. So what should I eat?

The most honest answer is first of all a statement about the state of knowledge. A review from Poznań puts it like this: so far there is no specific diet recommended for Hashimoto's [Review]. For a gluten free diet for everyone with Hashimoto's the evidence is not sufficient. What is presumed is a favourable influence of a low inflammation diet rich in vitamins and minerals with a smaller share of animal foods.

That is not a statement about the pointlessness of diet. It is a statement about how little has been studied systematically.

How little is shown by the systematic review of 2023 in a single number: out of 1,350 publications, nine evaluable papers remained [Systematic Review]. Three on gluten elimination, one on lactose elimination, two on energy restriction with or without exclusion of individual foods, two on black cumin and one on iodine restriction. Duration: 21 days to 12 months. Nine papers for a topic on which tens of thousands of pages have been written in the German language internet.

This is the point I make for the second time in this article, because it appears in no German source and because it puts everything else into perspective. When someone drops grain, milk and sugar for twelve weeks, the calorie amount, the meal structure, the alcohol consumption and often the sleep almost always change as well. Which of these made the difference, nobody knows. On the question of why calorie amount and calorie quality are not the same thing, I write in detail in Why a calorie in the body is not a calorie.

The basic pattern that emerges in the observational data

Together with the finding from Ülker that the Mediterranean group gained most clearly in free T4, a direction emerges. Not a list of prohibitions, but a pattern: plenty of vegetables, legumes, fruit, nuts, olive oil, fish, little highly processed food. That is unspectacular. It is also the only thing that can be derived halfway consistently from two independent data sources.

Selenium and zinc through food, and the most open contradiction of this topic

Selenium and zinc are the two trace elements that are used directly in the thyroid: selenium for the selenoproteins that work in the conversion of T4 to T3 and in antioxidant protection, zinc as a cofactor at several places in the same system. Through food they come above all from fish and seafood, meat, eggs, legumes, whole grains, nuts and seeds. Brazil nuts are the best known selenium source, and at the same time the one with the greatest range of variation, because the content depends on the soil.

I deliberately name no dosages here. That belongs in a medical conversation with a measured status, and the details are covered in Selenium, zinc, iron and vitamin D. What belongs here is a contradiction that I consider the most honest available information on the subject of selenium.

For vitamin D the data look somewhat friendlier, though exclusively as a supplement and not as a dietary question. A meta-analysis of 12 studies with 862 people found TPO and Tg antibodies clearly lower [Meta-analysis, k=12, n=862]. A second, more cautious paper covering 6 studies with 258 people confirmed the direction for the TPO antibody, but found no significant differences in TSH, FT3 and FT4 [Meta-analysis, k=6, n=258]. So antibodies move, while thyroid function does not reliably follow. Details, forms and measurement belong in the micronutrient article of the cluster.

What diet cannot do

The guideline of the American Thyroid Association on the treatment of hypothyroidism from 2014 systematically worked through 24 questions and arrives at a clear result: levothyroxine remains the treatment standard [Guideline]. For the alternatives examined, the working group found no consistently strong evidence of superiority for health endpoints.

For this article that means three things, and I would rather say them too clearly than too quietly.

First: diet does not replace replacement therapy. In none of the 32 sources evaluated here did a hormone dose become unnecessary through a dietary change. Second: Hashimoto's is a chronic autoimmune condition. There is currently no dietary pattern with a documented effect on thyroid function or on the course of the condition itself. Third: what diet can influence is wellbeing, inflammatory burden, the supply of building blocks and the absorption of the medication. For some people that is a great deal. It is just something different from what is written on the sales pages.

If you continue to feel unwell despite well adjusted values, that is no reason to leave out even more. It is a reason to keep looking. What can lie behind this constellation I have described in Normal values, symptoms anyway and in Functional hypothyroidism.

What I observe in the consultation

Here ends what studies carry. What comes now is observation and is explicitly marked as such.

I notice that people with Hashimoto's often come with very long lists of prohibitions and with very short answers to two simple questions: when do you take your tablet, and what do you drink with it. I also often see that ferritin, vitamin D, vitamin B12 and selenium have never been measured, while four food groups have been cut at the same time. And I regularly encounter the situation that a celiac work up is no longer cleanly possible after two years of a gluten free diet.

These are patterns, not study results. I describe them because they can be remedied without anyone having to leave anything out.

Frequently asked questions about diet with Hashimoto's

Where this topic goes on

Diet with Hashimoto's is a crossroads, not a dead end. Four roads lead on from here, and all four answer a question that this article deliberately only touched on.