Raw milk is the rare wellness claim that fails on every axis at once — safety, nutrition, and its own internal logic. That last one is the interesting part, so we will spend most of our time there.
The safety question is not close
Between 1998 and 2018, the FDA counts 202 outbreaks linked to drinking raw milk, causing 2,645 illnesses and 228 hospitalizations. A peer-reviewed analysis of 2013–2018 found 75 outbreaks and 675 illnesses, of which 48% were in people aged 0–19 — this is disproportionately a children's illness.
The number that matters most is a modelling study in Emerging Infectious Diseases. Raw milk drinkers are a small slice of the population — about 3.2% for milk, 1.6% for cheese — but per unit consumed they are roughly 840 times more likely to get an outbreak-related illness and 45 times more likely to be hospitalized than people drinking pasteurized dairy.
That figure is a model, not a headcount, and it covers four pathogens. It is also so large that the modelling assumptions would have to be catastrophically wrong to change the conclusion.
One honest caveat, since raw milk advocates raise it and they are right to: the same paper found Listeria illnesses were more often associated with pasteurized cheese. That reflects contamination after pasteurization, and amounts to roughly one additional illness a year. It does not offset an 840-fold ratio.
The nutrition question isn't close either
The claim is that heat destroys milk's nutritional value. Measured, it mostly doesn't.
Protein digestibility is 80.2% raw versus 80.0% pasteurized — a rounding error. Calcium absorption and bone deposition show no difference between raw, HTST and UHT milk. Fat composition is essentially unaffected. Bovine IgG is heat-stable enough that standard pasteurization denatures about 1% of it.
The one genuinely heat-sensitive nutrient is vitamin C, reduced by around 15%. Milk contains about 5 mg per cup — it is not a meaningful source of vitamin C either way. One study found vitamin C actually keeps better during storage in heated milk.
Now the enzymes, which is where the claim eats itself
This is the argument's centrepiece: pasteurization destroys living enzymes. Follow it through and it collapses in four separate places.
The main enzyme survives anyway. Plasmin, milk's principal proteolytic enzyme, is heat-stable. A large proportion survives pasteurization — and 30–40% of its activity survives even UHT treatment, which is far harsher.
The enzyme that is destroyed does nothing for you. Pasteurization does inactivate lipoprotein lipase. Lipoprotein lipase has no role in human digestion of milk fat; your own gastric and pancreatic lipases do that job. Its destruction is actually desirable, because residual activity turns milk rancid.
Enzymes are proteins, and you digest proteins. Any enzyme that did survive into your stomach meets acid and proteases and is denatured and broken down, exactly like every other protein you eat. This is the part of the argument that requires forgetting how digestion works.
The enzyme that would genuinely help isn't in cow's milk at all. Bile-salt stimulated lipase meaningfully aids fat absorption — and it is a human-milk enzyme, absent from bovine milk. There is also no lactase in milk, so raw milk cannot relieve lactose intolerance.
And a detail that inverts the whole frame: high enzyme activity in raw milk is a warning sign. Elevated protease activity indicates either mastitis — milk from an infected udder — or bacterial counts in the millions per millilitre.
"Beneficial bacteria"
Probiotics are, by definition, non-pathogenic organisms of human origin. The bacteria in raw milk arrive from infected udder tissue, soil, water, cow manure and milking equipment. They are not probiotic, and their presence is not a feature.
The FDA's note on bifidobacteria is the cleanest version of this: properly collected raw milk should not contain them, because their presence indicates faecal contamination.
And then bird flu happened
Since March 2024, H5N1 has been circulating in US dairy cattle, and infected cows shed enormous quantities of virus into milk — titres up to 10^8.8 TCID50 per millilitre. Refrigeration preserves it: the virus's half-life in raw milk is about 2.1 days at 4°C.
In November and December 2024, California regulators found H5N1 in retail raw milk, and Raw Farm LLC recalled its raw whole milk and cream, with the farm placed under quarantine.
Pasteurization inactivates it. Infectious virus has a half-life of 4.5 seconds at 63°C. The strongest real-world evidence is the FDA's retail survey: of 297 samples, 20% carried detectable viral RNA, and none contained viable virus. Fragments of dead virus are what successful pasteurization looks like.
A note on how that evidence gets misused: a study finding residual virus at 72°C for 15 seconds circulates as proof pasteurization fails. Its authors explicitly warned their spiked-milk lab setup does not replicate commercial pasteurizers. The retail survey settles the practical question.
The claim's real argument
"Thousands of years of tradition" is doing the persuasive work here, and it is worth answering directly rather than sneering at it. Those thousands of years also included bovine tuberculosis, brucellosis, and childhood deaths from milk-borne infection so routine they were unremarkable. Pasteurization is among the most successful public health interventions ever deployed. Treating it as an act of tampering requires forgetting what it was a response to.