Dairy: healthy, poorly tolerated or both at once

It is quarter past seven. The coffee is ready, the milk is standing next to it. And the same small debate as yesterday starts up in your head.

One voice says: milk is calcium, protein, an ordinary food. The other says: cow's milk is for calves, it gives you pimples and a bloated belly. Both sound certain, both point to studies.

Hardly any topic comes up as often in my consultations. And hardly any is sorted so badly. Because behind the harmless sentence “I do not tolerate milk” sit three completely different processes in the body.

I will lay those three processes out for you. Then comes the part almost every guide leaves out: milk, yoghurt and cheese behave differently in the same studies, sometimes even in opposite directions. At the end there is no verdict, but a route to your own answer.

Why milk of all foods is so contested

Ask yourself a question you otherwise never ask: who was this food actually made for?

With an apple the answer is unclear. With milk it is obvious. Milk is the only food a mammal produces specifically for its own offspring, in a narrowly limited phase of life. A lot about it is built for growth. More on that in the text about breast milk, growth and hormones.

That many adults still split milk sugar is not a given. It is a young adaptation, and it is spread very unevenly around the world.

Modelling based on ancient genomes places the time frame within the last 10,000 years or so. Tolerating milk as an adult is therefore younger than farming across much of Europe. A large synthesis of all phenotype and genotype data also shows this: across much of Africa, the Middle East and Central Asia the known gene variants do not fully explain tolerance.

And now you know why the discussion about milk runs so hot. People with very different biology are arguing about the same drink.

Three questions that constantly melt into one

“I cannot tolerate milk.” When I hear that sentence, I ask back: what exactly happens, and after what?

Because three different processes carry the same everyday name. They involve different systems, need different tests and lead to different consequences.

Without that separation you end up testing the wrong thing. Someone with milk protein symptoms who takes a lactose breath test hears: everything is fine. And if they then eat lactose free while the protein was the trigger, they wonder why nothing improves.

With the immune question, the gap between suspicion and confirmation is even larger. In the European EuroPrevall survey, self-reported food allergy ranged from 6.5 to 24.6 percent depending on the centre, while probable allergy ranged only from 1.9 to 5.6 percent. In two German birth cohorts with 2,145 children, parents reported symptoms in 13.2 percent of one-year-olds. Medically diagnosed were 2.4 percent, and cow's milk sat right at the top of the list of suspects.

And now you know why two people with the same sentence on their lips may need completely different answers.

Lactose is a question of amount, not of yes or no

“One sip of milk and my belly is wrecked.” I hear that sentence often. It describes a real experience. Under controlled testing it usually does not hold up all the same.

Two years later the same group doubled the amount: twice 240 ml daily with meals. Here too, complaints under regular milk did not rise significantly. The group with the strongest self-attribution reported more gas symptoms in both phases, so also under lactose-free milk. The authors read that as a pre-existing tendency to bloat that was being attributed to lactose.

And now you know why “I cannot tolerate lactose” almost never means that no lactose at all is possible.

A1 versus A2 and the peptide beta-casomorphin-7

For some years now there has been a milk in the chilled aisle that costs almost twice as much and promises to be easier on the gut. Marketing or biochemistry? A bit of both. I am writing this section especially carefully, because it is the trickiest part of the whole text.

Beta-casein is one of the main proteins in cow's milk. At position 67 of the protein chain the A1 variant carries histidine, the A2 variant proline. A single building block of difference, and it changes how easily the chain breaks at that spot. Picture a string of pearls with a predetermined breaking point: on one it snaps easily, on the other barely at all. From A1 a fragment of seven amino acids can be released, beta-casomorphin-7. It can dock onto opioid receptors, and those sit among other places in the intestinal wall.

A crossover study with 45 participants, an investigation in preschool children, a US group with 33 people and an MRI study in ten people all point in the same direction. Now comes the other side, and it matters just as much.

A Korean crossover study in 40 people delivered opposite results in 2024 depending on the measuring instrument. And the most recent paper, a randomised pilot study from 2026, found no significant group difference at all between A1-free and conventional milk: not in symptoms, not in transit time, not in IgE, IgG or interleukin-4.

And now you know why two serious articles on A2 milk can reach opposite conclusions without either of them fudging anything.

Yoghurt and cheese are their own case

In almost every nutrition guide there is a single word: dairy. As if milk, yoghurt and cheese were the same thing in a different consistency.

In the data they are not. Within the same cohorts they sometimes move in opposite directions. That is one of the most striking findings in this whole topic, and it is told surprisingly rarely.

In metabolism the pattern continues. A dose-response meta-analysis across 22 cohorts with almost 580,000 people found only a weak inverse association for dairy overall, but for 80 grams of yoghurt daily a relative risk of 0.86 for type 2 diabetes. An umbrella review confirmed the direction in 2022. Observational data with high heterogeneity, which prove no cause, but do show one thing: in the metabolic field milk stands better than its reputation online.

The matrix idea, with a real experiment

The idea is called the food matrix: what counts is not the single nutrient but the structure it is packaged in. I explain the same idea using fries and banana.

In fairness: there are data pointing the other way. In an analysis with 46 people, butter raised HDL mediated cholesterol efflux capacity compared with cheese, measured in a cell assay and therefore far from everyday life. And four meals with 45 grams of milk fat each from butter, cheese, whipping cream or sour cream produced four different triglyceride curves in 47 adults. The core holds: structure beats the nutrition table, which fits the text about blood sugar spikes.

Ripened cheese is also one of the few relevant sources of long chain menaquinones, meaning vitamin K2 as MK-8 and MK-9. Online that has grown into a big story, while the data are smaller. A Dutch cohort with a good 16,000 women found a hazard ratio of 0.91 per 10 micrograms daily, with a confidence interval that touched 1.00. A larger cohort from the same country found no association.

Yoghurt and cheese are protein sources too, and how much you need is covered in the guide How much protein do you really need. Fermented food and the microbiome belong to the topic of fibre.

And now you know why a blanket verdict on “dairy” almost inevitably misses the point.

Bones and osteoporosis, where the evidence gets surprisingly thin

“Drink milk, it is good for your bones.” You probably know that sentence from childhood. It sounds so self-evident that hardly anyone asks back.

Let me say up front where this section does not lead: not to the claim that milk harms your bones. The data support that just as little.

Reverse causation here means: people who are frail or have an osteoporosis diagnosis are often advised to drink more milk. More frequent fractures afterwards are then not necessarily down to the milk. In the meta-analyses the picture also splits by product.

The largest of these analyses covered almost 487,000 people. Its authors point out themselves that the yoghurt and cheese studies were not adjusted for socioeconomic status. On top of that comes a geographical split: in the USA hip fracture risk fell to 0.75 with higher milk intake, while in Scandinavia it sat at 1.00. As an explanation the authors suspect vitamin D fortification in the USA.

And now you know why I neither recommend milk here nor advise against it, but ask the question differently.

Acne, IGF-1 and the skimmed milk paradox

There is an observation many young people make themselves, long before they read about it: leave milk out, skin calms down. Start again, skin gets restless again.

I am deliberately keeping this section short. The mechanics of acne, meaning sebaceous gland, androgens, insulin and skin barrier, are covered in detail in the text on hormonal acne from within. Here it is only about the milk-specific numbers. Three independent meta-analyses find an association, with odds ratios roughly between 1.2 and 1.5. What stands out is not the size, but the order.

The lowest fat variant of all is the most strongly associated one. That argues against milk fat as the trigger and turns attention towards protein and hormone signals. Two of the three analyses found no association or only a weak one for cheese and yoghurt, the same pattern as with bones. One analysis did however find an association for yoghurt. I am leaving that contradiction standing, because it is real.

The only randomised building block in this chain is the IGF-1 rise from the bone meta-analysis further up. The step from IGF-1 to acne remains observation, and to this day I cannot find a randomised elimination trial on it in the literature.

And now you know why the skimmed milk result is so interesting, even though the effect stays small.

Raw milk, and why I am cautious here

There is one finding in this topic that fascinates me professionally and that I still do not translate into a recommendation. That happens rarely, so let me explain it more fully.

A meta-analysis across eight studies confirmed the pattern with an odds ratio of 0.58 for asthma. And then those very authors write a sentence that is often overlooked: because of the minimal but real risk of life-threatening infections, they explicitly advise against consuming raw milk. Instead they point to an ongoing study using microbiologically safe, minimally processed milk.

One damper belongs here. In a US survey of more than 3,300 people, the strongest association with less atopy came not from the milk but from the mother's agricultural work during pregnancy. The farm effect is evidently a bundle.

And now you know why I am fascinated and reserved at the same time here.

Testing instead of believing: elimination and reintroduction

If you have read this far, you can sense where this is going. There is no general answer to whether milk is healthy. There is only an answer to what a particular product in a particular amount does in you.

That question can be answered, not by reading but through a cleanly built trial. Symptoms are diagnostically blurry, and suspicion turns up five to ten times more often than confirmation. If neither symptom nor self-assessment is reliable, you need a comparison condition. And that is exactly what you build with an elimination phase and a subsequent reintroduction.

Two things often get forgotten along the way. First: if no clear picture emerges, a different axis may be behind it. Ripened cheese, for example, is also high in histamine, which fits the text on histamine intolerance and DAO deficiency. Methodologically related is the approach from the 14-day glucose sensor.

Second: if the test turns into a permanent solution, it becomes a question of calcium, iodine, vitamin B12 and protein quality. I work through that calculation in the text on vegan nutrition. Dropping dairy in general is not a recommendation I make across the board. How much processing counts belongs in the picture too, which is where unprocessed food and satiety and the calorie myth fit in.

And now you know why I do not end this article with a recommendation, but with a method.

Common questions about milk and dairy

Milk in the larger context

The question of tolerance rarely ends with milk. It leads to the skin, to histamine metabolism, to the fermentable carbohydrates in the large intestine and to the fat soluble vitamins. Several routes lead onward from here.