Fermented Foods

Yogurt and the other cultured milks track with slightly lower death rates, and natto, the Japanese fermented soybean, has the biggest single-food number in the data, from the narrowest evidence. Sauerkraut, kimchi and kombucha have interesting gut and immune data but no mortality signal behind them. And the live bacteria on the label matter much less than the marketing says.

Aim for about two servings of fermented food a day, varied across types — a pot of plain yogurt or a glass of kefir, a few forkfuls of sauerkraut or kimchi, a pack of natto if you like it. Eat them daily, alongside a fibre-rich diet, and build up over a week or two. If you take warfarin or another blood thinner, or an older-style antidepressant called an MAOI, read the safety section before you start.

Fermented dairy: a real signal, smaller than advertised

The most consistent result in this whole field is for fermented milk. Pooling prospective cohorts, people who ate more of it had about 6% lower all-cause mortality, 7% lower cardiovascular mortality and 6% lower cancer mortality. For yogurt on its own, the link with heart deaths was weaker and not statistically significant. Aged cheese, analysed separately, came out at about 3% lower all-cause mortality and nothing for heart disease or cancer. Kefir had no pooled estimate at all.[1] These are single-digit differences from observational data, not dramatic effects.

There is a good reason not to give fermentation all the credit. A pooled analysis of ten Japanese cohorts — nearly 400,000 adults and about 70,000 deaths — found no association for yogurt or cheese in either sex, while ordinary unfermented milk tracked with about 6% lower mortality in men.[2] Whether fermentation or dairy is doing the work is open.

Two other signals are worth knowing:

  • Diabetes. Across 22 cohorts and nearly 580,000 people, eating about 80 g of yogurt a day went with roughly 14% lower risk of type 2 diabetes compared with none.[3] In 2024 the US Food and Drug Administration let yogurt makers say that at least 2 cups a week "may reduce the risk of type 2 diabetes", noting the evidence is limited. It followed an industry petition and rests on the same observational data, so it confirms the direction, not the cause.[4]
  • Blood pressure. In 14 randomised trials, probiotic fermented milk lowered blood pressure by about 3 mmHg systolic and 1 mmHg diastolic.[5]

Kefir, a more complex culture than yogurt, has weaker human data than claimed. Insulin-resistance scores often fall in kefir trials, but the one trial against the same amount of plain milk, in people with metabolic syndrome, found no significant difference in the 22 analysed for this report, too few to rule one out,[6] and a review of the kefir trials calls the overall evidence "limited".[7] Any blood-sugar benefit is inconsistent and not clearly a fermentation effect. The memory and mood claims are weaker still: pooling 22 trials of prebiotics, probiotics and fermented foods found no significant effect on any area of thinking, and the mood findings come from small single trials.[8] Coconut and water kefir are fine lactose-free swaps, but untested as equivalents.

Natto: the biggest number, the narrowest evidence

Natto carries an order of magnitude more of the vitamin K2 form called MK-7 than any other common food, and habitual eaters have modestly higher bone density across four Japanese cohorts.[9] The "keeps calcium out of your arteries" half of the story rests on K2 supplement trials, which are split. See Vitamin K2.

In 1,548 older Japanese men followed for about 12 years, eating natto several packs a week went with about 40% lower all-cause mortality, an association that weakened and stopped being significant at a pack a day or more.[10] In about 29,000 Japanese adults over 16 years, the highest natto eaters had about 25% lower cardiovascular mortality and roughly a third lower stroke mortality than the lowest.[11] Both are observational, in Japanese populations.

The rest of the natto story is thinner. Its clot-dissolving enzyme, nattokinase, did nothing for artery disease or any blood measure in the one long trial, 265 adults over a median three years,[12] and the anti-ageing claims rest on a roundworm study, with no rodent or human data.

If you can't stand the taste, a daily MK-7 supplement of 90–180 µg covers the vitamin K2, but not the enzyme or anything else in the food.

Why live cultures matter less than you'd think

The old idea was that fermented foods work by planting their bacteria in your gut. That has not held up: the adult gut is resilient, and swallowed Lactobacillus species mostly pass through, with any rise in their numbers fading.[13] What seems to matter more is the whole food — acids and other compounds the microbes have already made, fragments of dead bacteria that the gut's immune system responds to whether or not they arrive alive, and partly digested fibre that feeds your own gut bacteria.

The cleanest test of the dead-bacteria idea is not a food trial. 200 adults with diarrhoea-type irritable bowel syndrome got a live bacterial strain, the same strain heat-killed, or placebo, and the dead version worked almost exactly as well as the live one. The trial was industry-funded, with three authors employed by the sponsor.[14]

The best food trial gave 87 adults 100 g a day of fresh, live sauerkraut and of pasteurised sauerkraut, four weeks each. Each version shifted a few different species; only the pasteurised one raised short-chain fatty acids in the blood, in an analysis not corrected for multiple testing. The trial never compared the two versions directly, and its own conclusion is that neither had any overall impact on the gut microbiome of healthy people.[15] So a pasteurised supermarket sauerkraut is not the dead, useless product live-culture marketing implies — but a fresh, refrigerated, unpasteurised one is still the better default for variety.

A Stanford trial of a high-fermented-food diet raised gut diversity and lowered 19 of 93 inflammatory signalling proteins over ten weeks. But its pre-registered main outcome did not change, each group had only 18 people, and participants ended up on about six servings a day — roughly triple the target above.[16]

Sauerkraut and kimchi: gut shifts, no mortality signal, and a salt caveat

Vegetable ferments have weeks of microbiome data and nothing on lifespan. The two sauerkraut studies — the crossover above, and a ten-day study of ten athletes eating 250 g a day with no control group — found modest shifts in individual species. Only the athlete study checked inflammation, insulin and blood sugar, and none of them changed; the crossover measured none of these.[17][18] The one cohort to look at fermented foods by type, about 34,000 Dutch adults over 15 years, found no link between total fermented-food intake and mortality.[19]

Kimchi has intriguing immune data. In a 12-week double-blind trial of 90 overweight adults, a detailed analysis of blood immune cells — in just thirteen of them, four or five per group — suggested kimchi powder primed both the immune system's defensive readiness and its anti-inflammatory restraint.[20] The dose was 3 g a day of freeze-dried powder in capsules, not kimchi as food, and nothing shows it prevents any disease.

The real caveat is stomach cancer. Across prospective cohorts, gastric-cancer risk rose about 15% for every extra 40 g a day of pickled vegetables, driven by salt and by nitrites that form cancer-causing compounds.[21] For scale: Korean adults average about 128 g of kimchi a day, against about 11 g of pickled vegetables in Japan, and the Korean cohorts showed no dose-response at all because even their lowest eaters ate more than most populations' highest. In a large Korean survey of about 116,000 adults, 1–3 servings a day went with 10–12% less obesity in men only — and even there the overall trend wasn't significant — with the highest intakes trending the wrong way; in women, two-thirds of the sample, there was no link.[22] The risk shows most clearly at high intake; moderate servings haven't been shown harmful, though no safe threshold has been established either.

Buy brine-fermented cabbage: vinegar-pickled "kraut" never fermented, so it has none of the microbial products.

Kombucha: thinner than its reputation

Kombucha's blood-sugar reputation outruns its trials. The field's systematic review — eight trials, 303 adults — found "minimal effects, including inconsistent findings on glucose metabolism", and called the clinical data inconclusive.[23] The famous diabetes number, fasting glucose falling from about 164 to 116 mg/dL, is not a result: kombucha did not beat placebo in that trial, and the figure rests on the seven people who finished, measuring their own glucose at home.[24]

Many commercial kombuchas carry 2–8 g of sugar per 250 mL, with flavoured ones at the top, so read the label. And kombucha is mildly alcoholic: US regulators treat it as an alcoholic drink if it reaches 0.5% alcohol at any point, including by fermenting on in the bottle,[25] and a batch used in one clinical trial measured 1.5%.[26] Worth knowing in pregnancy, or if you avoid alcohol.

How to actually eat them

What counts as a serving. Stanford's nutrition group publishes a working guide, noting there are no official guidelines or scientific consensus yet. Its servings are by volume; the gram figures and the natto row below are added for convenience and are not Stanford's.[27]

FoodOne serving
Sauerkraut, kimchi, traditional pickles¼ cup, ~35–50 g
Yogurt, cottage cheese, kefir~175 mL
Kombucha, water kefir~175 mL
Miso1 tbsp (~17 g)
Natto~40 g (one pack)

Two a day is a floor, not a proven optimum. That is the minimum Stanford's guide suggests, "and more as tolerated".

Build up slowly if your diet is low in fibre. Jumping straight to two servings can bring a week of bloating and gas, though tolerance varies: the sauerkraut trial started 87 adults on 100 g a day with no ramp-up, and 90–95% were content with no change in bowel habit.[28] A sensible rule of thumb is about 50 g of sauerkraut or 50 mL of kefir, built up over 7–14 days, before adding a second type.

Pair them with fibre. What your own gut bacteria can do with a ferment depends on what they have to eat. Vegetables, legumes, whole grains and fruit come first; ferments add to that diet, not replace it.

Daily, not as a course. The bacteria a ferment adds disappear again within a few weeks of stopping.[29]

Fermenting at home. Fast souring is what keeps a ferment safe. With 2–2.5% salt by weight, acidity should start rising after two to three days and pass the level at which botulism bacteria cannot make toxin (pH 4.6) by the end of the first week. Throw out any batch that is still not sour, or is slimy or mouldy, at a week.[30] Properly soured vegetables are too acidic to carry botulism, but home fermentation isn't exempt: in US surveillance, home-prepared foods other than home-canned ones are the largest single source of foodborne botulism, about a third of it fermented fish and marine mammal. Illness from fermented vegetables has come from E. coli carried in on the raw produce. Raw-milk dairy is a separate and larger risk.

Who should be careful

  • Anyone on a blood thinner — warfarin, a direct oral anticoagulant, aspirin or clopidogrel. Natto is the problem, twice over. Its vitamin K2 directly opposes warfarin and the other vitamin-K-antagonist drugs (the newer direct oral anticoagulants are not expected to be affected). And nattokinase dissolves clots, and has never been tested alongside any of these drugs — untested, not shown safe. Talk to your prescriber before making natto a daily habit. See Vitamin K2.
  • Anyone taking a classic MAOI antidepressant — phenelzine, tranylcypromine, isocarboxazid, or high-dose selegiline. With these drugs, tyramine in food can push blood pressure up sharply, and the ferments richest in it are aged cheese, fermented soy, and fermented sausage and fish. Modern production has cut tyramine in most foods, so the restriction is narrower than old diet sheets imply, but clear this one with a prescriber.[31]
  • Histamine intolerance or mast cell activation syndrome. Ferments can trigger flares, but histamine varies more within a food than between foods: fermented soy averages about 74 mg/kg; aged cheese about 33 mg/kg but up to 390; sauerkraut about 44 mg/kg, so roughly 2 mg in a ¼-cup serving and up to about 4 mg.[32] Yogurt is essentially histamine-free; kombucha hasn't been measured. If you react, find out which foods and batches rather than cutting the whole category.
  • Active small intestinal bacterial overgrowth (SIBO) or severe bloating with irritable bowel syndrome. Ferments are rich in fast-fermenting material and can trigger flares. Test one at a time; some people tolerate yogurt and miso but react to kombucha and sauerkraut.
  • Poorly controlled blood pressure — count the sodium. A ¼-cup of sauerkraut carries roughly 230–330 mg and kimchi 175–250 mg, about a tenth of a day's allowance or a little more; traditional dill pickles about 280–400 mg. Miso is the one to watch: about 630 mg per tablespoon.
  • A weakened immune system. The documented risk is specific: soft cheese made from unpasteurised milk carries a listeria risk per serving 50 to 160 times that of the pasteurised version.[33] Vegetable ferments carry the ordinary raw-produce risk, and the bloodstream and fungal infections reported in this group come mostly from concentrated probiotic supplements, not food. Check with your treating team.

What's overhyped

  • "Reverses biological age by years." The headline traces to an 8-week pilot in 43 men aged 50–72, 38 of them analysed, that reported a 3.2-year difference on an ageing clock,[34] and an uncontrolled series of six women reporting 4.6 years.[35] Neither programme included any fermented food. The pilot has since carried two published corrections, and the clock it used varies by around three years between repeat measurements of the same sample — about the size of the claimed effect.
  • Sourdough as a longevity food. The pooled analysis found no significant link between bread and mortality, though from only two studies and none on sourdough specifically.[36] Baking kills the culture outright. Good bread, not a longevity food.
  • High bacteria counts on labels. Live bacteria survive stomach acid poorly, and even a "10 billion CFU" product works mostly through dead-cell fragments and fibre. Probiotic capsules are a separate category; the evidence here is about daily food.
  • Ginger bug, jun and other home-brews as probiotic elixirs. The microbes in home-brewed ginger beer have barely been studied — two sequencing studies worldwide — and the one survey of a home batch found it dominated by environmental bacteria of unknown relevance, not the familiar gut-friendly kinds. The drinks are fine; the probiotic claim is not.
  • Sweetened, flavoured supermarket "ferments" don't necessarily carry the benefits of traditional ones: sugar and processing matter. Probiotic-fortified cereals, gummies and snack bars are processed foods first, with fermentation as a marketing layer. See ultra-processed food.

Further reading

  • Matalas A, et al. Fermented foods consumption, all-cause, and cause-specific mortality: a meta-analysis of prospective cohort studies. Front Nutr 2026.[37]
  • Wastyk HC, Fragiadakis GK, Perelman D, et al. Gut-microbiota-targeted diets modulate human immune status. Cell 2021.[38]
  • Gijsbers L, Ding EL, Malik VS, et al. Consumption of dairy foods and diabetes incidence: a dose-response meta-analysis of observational studies. Am J Clin Nutr 2016.[39]
  • Nagata C, Wada K, Tamura T, et al. Dietary soy and natto intake and cardiovascular disease mortality in Japanese adults: the Takayama study. Am J Clin Nutr 2017.[40]
  • Fujita Y, Kouda K, Tamaki J, et al. A 15-year cohort study of self-reported fermented soybean (natto) intake and all-cause mortality in elderly men. Clin Nutr ESPEN 2025.[41]
  • Yoo JY, Cho HJ, Moon S, et al. Pickled vegetable and salted fish intake and the risk of gastric cancer. Cancers 2020.[42]
  • Marx W, Scholey A, Firth J, et al. Prebiotics, probiotics, fermented foods and cognitive outcomes: a meta-analysis of RCTs. Neurosci Biobehav Rev 2020.[43]
  • Schropp N, et al. The impact of regular sauerkraut consumption on the human gut microbiota: a crossover intervention trial. Microbiome 2025.[44]
  • Bellikci-Koyu E, Sarer-Yurekli BP, Akyon Y, et al. Effects of regular kefir consumption on gut microbiota in patients with metabolic syndrome: a parallel-group, randomized, controlled study. Nutrients 2019.[45]
  • Praagman J, et al. The relationship between fermented food intake and mortality risk in the European Prospective Investigation into Cancer and Nutrition-Netherlands cohort. Br J Nutr 2015.[46]
  • Mendelson C, Sparkes S, Merenstein DJ, et al. Kombucha tea as an anti-hyperglycemic agent in humans with diabetes — a randomized controlled pilot investigation. Front Nutr 2023.[47]
  • Leeuwendaal NK, et al. Fermented foods, health and the gut microbiome. Nutrients 2022.[48]
  • What counts as a serving of fermented foods? — Stanford Medicine clinical-nutrition guidance.[49]
— § —