How milk kefir fermentation works: the culture eats the lactose and produces acid. The pH falls from about 6.7 to about 4.2 to 4.6 in 24 hours, and that acidity makes the casein coagulate, which is what turns milk creamy. Fat, protein and calcium stay as the milk had them.
The figures on this page
How milk kefir fermentation works comes down to one thing. The culture eats the lactose in the milk and produces acids. Everything else follows from that.
The lactose is split and used. Lactic acid bacteria turn it mainly into lactic acid. Yeasts, which are also part of the culture, produce a little carbon dioxide and traces of alcohol.
The acidity rises. The pH falls from about 6.7, which is milk, to about 4.2 to 4.6 after 24 hours.
The casein coagulates. Milk protein changes structure as the pH drops, and that is what turns a liquid into something creamy. It is the same mechanism as in a yoghurt, at a lower temperature.
And past a point, it separates. Keep going and the coagulated protein contracts and squeezes out the whey, which is the layer you see at 36 hours.
This is the part that explains the nutrition figures, and it is more useful than any table.
Changing: the lactose falls, the acidity rises, the texture thickens, and traces of alcohol and a little carbon dioxide appear.
Not changing in any meaningful way: the fat, the amount of protein and the calcium. They are the milk’s, and fermentation barely touches them.
Which is why a 200 ml glass from whole milk sits at 110 to 130 kcal, about where a glass of milk sits. Anyone wanting to lower that changes the milk, not the culture.
10 to 20 g of grains per litre of milk, no sugar, 24 hours at 20 to 25 °C. The lactose in the milk is the food, which is why no sugar is added.
| What goes in | Water kefir, 1 litre | Milk kefir, 1 litre |
|---|---|---|
| Culture | 30 g of grains | 10 to 20 g of grains |
| Liquid | 1 litre of unchlorinated water | 1 litre of milk |
| Sugar | 4 tablespoons, about 60 g | none, the lactose is enough |
| Dried fruit | 1 fig or 2 dried apricots | none |
| Citrus | 1 lemon or 1 orange | none |
| Duration | 24 h, up to 48 h | 24 h |
| Temperature | 20 to 25 °C | 20 to 25 °C |
Understanding how milk kefir fermentation works is mostly useful because it tells you which dial to turn when the result is wrong.
Temperature, first. At 20 °C a cycle takes a full 24 hours or a little more. At 26 °C it is closer to 18. Below 18 °C it slows markedly. The culture is not damaged by a cool kitchen, it just works slower.
Quantity of grains, second. 10 g per litre gives a slower, milder result; 20 g a faster, sharper one. This is the dial most people never touch, and it is the one that fixes a result that is persistently too sour.
The milk, third. UHT milk often gives a thinner result because the heat treatment has already altered the proteins. Skimmed milk gives something thinner and sharper. Raw milk gives a fuller result but brings its own microorganisms, so it varies more between cycles.
Age of the grains and jar size matter least, but a jar filled to the brim leaves the small amount of gas nowhere to go and will lift a tight lid.
Milk kefir has a narrower window than water kefir, and the mechanism explains why.
At 24 hours the casein has coagulated enough to be creamy but not enough to contract. The drink pours, tastes tart, and holds together.
At 36 hours the coagulated protein has contracted and expelled the whey. You get a clear layer at the bottom and a grainy layer on top. Nothing has gone wrong and nothing is unsafe, but it is no longer a drink.
Both parts are usable: the whey in a dough that will be baked or in a marinade, the curd strained in a cloth as a spreadable fresh cheese. That is the same coagulation taken one step further, deliberately.
And why it curdles when heated. Heat makes the already acidified casein contract fast and completely. A kefir added to a boiling soup separates in seconds. It always goes in at the end, once the heat is gone.
These are orders of magnitude, not laboratory values. They follow from the recipe quantities and the fermentation time.
| Per 200 ml glass | Water kefir, 24 h | Milk kefir, 24 h |
|---|---|---|
| Energy | 30 to 50 kcal | 110 to 130 kcal |
| Residual sugar | about 3 to 4 g per 100 ml | little, lactose partly used |
| pH | about 3 to 3.5 | about 4.2 to 4.6 |
| Alcohol | 0.2 to 1 per cent | 0.2 to 1 per cent |
| Carbon dioxide | marked, more after bottling | little |
What is missing from this table is a column for an effect. It is not missing through carelessness. We could only invent it.
First, cutting the sugar. By far the most common. The 4 tablespoons per litre feed the grains, not the drinker. After 24 hours a good part is gone. Halving it starves the culture, and within three cycles it stops multiplying. For a less sweet drink, extend the time.
Second, chlorinated water. Chlorine is made to kill microorganisms, and it does the same to your culture. Bottled water, filtered water, or tap water left uncovered for 12 hours all solve it.
Third, a jar that is too small. A litre of drink needs a 1.5 litre jar. The carbon dioxide has to go somewhere. A jar filled to the brim overflows and smells unpleasant within two days.
Fourth, rinsing every cycle. The grains are rinsed only when visibly coated, in cool unchlorinated water. Rinsing at every turn takes away the film that protects them.
Fifth, metal. A plain metal sieve or spoon reacts with the acidity. Plastic or stainless steel solves it, and it is the only point where equipment genuinely matters.
The best way to understand how this works is to follow one jar from arrival to its steady rhythm.
Day 1, activation. The grains arrive hydrated. They go into 30 cl of water with 40 g of sugar for 24 hours, with no fig and no citrus. This is a wake-up cycle, not a drink. The liquid is poured away.
Days 2 to 7, the first cycles. The first two drinks may be flat or too sweet. That is not a fault in the grains, it is the time they need. Strain at 24 hours anyway, always with the same quantities.
Week 2, the rhythm. The drink becomes consistent. Bubbles at the rim, a slightly tart taste, the fig rising around the twelfth hour. From here you can tell by eye whether a jar is ready.
Weeks 3 to 6, multiplication. The grains visibly take up more room. When you strain, you have more than you put in. At that point you divide. Part stays in the jar, the rest starts a second litre or is dried as a reserve.
It is worth being explicit, because knowing the mechanism makes it tempting to go a step further than the evidence does.
It does not tell you the microorganism count. Knowing that bacteria and yeasts are at work is not knowing how many. A count needs a laboratory analysis of a specific batch, and temperature, milk, duration and grain age shift it enough that two jars in the same house differ.
It does not tell you which species. Species lists circulate widely, taken from studies on particular cultures, and they look far more precise than they are.
It does not tell you anything about a body. The chemistry of the jar is observable. What happens after you drink it is not something a culture seller measures, and no health claim about kefir is authorised in the European Union.
And it does not make raw milk safe. The pH falls to 4.2 to 4.6, which is acidic, but acidification is not pasteurisation. A culture does not turn raw milk into pasteurised milk.
Too sour. Strain earlier, at 20 to 22 hours, or drop to 10 g of grains per litre, or find a cooler spot. In that order.
Too thin. Almost always the milk. UHT and skimmed give thinner results. Switch to whole milk, or extend by two or three hours.
Separating too early. Too warm, or too many grains. Both make the coagulation run ahead of the clock.
Not thickening at all. Too cold, too few grains, or a culture still settling in after arrival. Give it three cycles before concluding anything.
Every one of those follows directly from how milk kefir fermentation works: acid production sets the pace, and the pace sets the texture. Once that is clear, the recipe stops being a set of instructions and becomes something you can steer.
The health claims authorised in the European Union are listed in the register of the European Food Safety Authority. Guidance on home food hygiene is published by the Food Standards Agency.
Live cultures, water and milk
Shipped hydrated and active, ready for the first cycle on the day they arrive.
See the culturesThe whole mechanism is one sentence. Acid production sets the pace, and the pace sets the texture. Everything practical about milk kefir follows from that.
And once that is clear, how milk kefir fermentation works stops being theory. Too sour means the acid ran ahead, too thin means the milk had less to coagulate, and separated means it ran further still. The recipe becomes something you can steer rather than follow.
The culture eats the lactose in the milk and produces mainly lactic acid. The pH falls from about 6.7 to about 4.2 to 4.6 in 24 hours, and that acidity makes the casein coagulate, which is what turns liquid milk into something creamy. Yeasts in the culture also produce a little carbon dioxide and traces of alcohol.
It changes the lactose, which is partly consumed, the acidity, the texture, and it adds traces of alcohol. It does not meaningfully change the fat, the amount of protein or the calcium. Those stay as the milk had them.
Because milk protein, casein, changes structure as the pH drops. It is the same mechanism as in a yoghurt, happening at 20 to 25 °C instead of 40 to 45 °C.
Because the coagulated protein keeps contracting and squeezes out the whey. You get a clear layer at the bottom and a grainy layer on top. Nothing has gone wrong. Both parts are usable.
Temperature first: 24 hours at 20 °C, closer to 18 hours at 26 °C. Then the quantity of grains: 10 g per litre is slower and milder, 20 g faster and sharper. Then the milk.
Because heat makes already acidified casein contract fast and completely. Added to a boiling soup it separates in seconds. It always goes in at the end of a preparation, once the heat is gone.
No. The pH falls to 4.2 to 4.6, which is acidic, but acidification is not pasteurisation. A culture does not turn raw milk into pasteurised milk.
Almost always the milk. UHT milk gives a thinner result because the heat treatment has already altered the proteins, and skimmed milk is thinner still. Switch to whole milk or extend by two or three hours.
In this order: strain earlier at 20 to 22 hours, drop to 10 g of grains per litre, or find a cooler spot. A consistently mild result usually comes from 10 g and 20 to 22 hours.
No. Knowing that bacteria and yeasts are at work is not knowing how many. A count needs a laboratory analysis of a specific batch, and a kitchen jar is not analysed.
