Choosing Milk for Cheese: What Actually Matters
Milk choice sets the ceiling on everything downstream. You can rescue a mediocre make with good technique, but you cannot make more cheese than the milk contains, and some milks will not make cheese at all.
Fat and casein set the yield
The yield calculator uses the Van Slyke formula, in which fat and casein appear in the numerator and target moisture in the denominator. Hold moisture at 50% and vary only the milk:
| Milk | Fat | Protein | Yield per 100 kg |
|---|---|---|---|
| Semi-skimmed cow | 1.5% | 3.3% | 8.43 kg |
| Whole cow | 3.6% | 3.2% | 12.52 kg |
| Sheep | 7.5% | 5.8% | 22.59 kg |
Semi-skimmed loses about a third of the yield against whole milk despite having slightly more protein — because fat is nearly as important as casein in the formula and there is far less of it. Sheep milk nearly doubles whole cow milk, which is the arithmetic behind the price of sheep-milk cheeses.
Goat milk sits closer to cow milk on yield but behaves differently in the vat: its curd is typically softer and more fragile, which is a handling difference rather than a quantity one.
The heat treatment matters more than the fat
Before comparing fat percentages, check the processing, because it decides whether you get a curd at all.
Pasteurised milk is the standard choice and works well. It may benefit from added calcium chloride, since pasteurisation affects the calcium balance that rennet coagulation depends on.
Ultra-pasteurised or UHT milk generally will not set a proper curd. The higher heat denatures whey proteins in a way that interferes with coagulation, and extra rennet does not fix it. This wastes more first attempts than any other single factor, and the carton is the only place to find out. Check it before you check anything else.
Raw milk can make excellent cheese and carries pathogen risk that pasteurisation exists to remove — Listeria, Salmonella, E. coli O157, Campylobacter. If you use it, sourcing and cold chain are the whole question, legal status varies by jurisdiction, and the highest-risk combination available is raw milk in a fresh, high-moisture cheese. That is a considered choice for an experienced maker, not a starting point.
Homogenisation is a smaller issue than people think
Homogenisation breaks up fat globules so they stay suspended. It does slightly change curd behaviour — the curd tends to be softer and can retain a little more moisture — but most home cheesemaking works perfectly well with homogenised milk, and non-homogenised is a preference rather than a requirement for most styles.
It matters most where you are deliberately separating cream, or making a style whose texture depends on intact fat globules.
Reading the carton, and its limits
Nutrition panels give fat and protein per serving, which converts to a percentage. Two cautions.
First, the panel prints total protein; the yield formula wants casein, which is roughly 78% of it in cow's milk. The calculator applies that conversion, so enter the carton figure and let it do the arithmetic rather than adjusting twice.
Second, printed figures are averages. Milk composition shifts seasonally — fat and protein are generally higher in late lactation and lower in flush — sometimes by enough to notice in a yield. Farm milk often comes with an actual testing report, which is genuinely better data. It says nothing about microbiological safety, which is a separate question entirely.
Cost per kilo of cheese, not per litre of milk
The comparison worth making is not what the milk costs but what the cheese costs, and richer milk can be cheaper on that basis even at a higher price per litre.
Run both milks through the yield calculator at your target moisture, divide the milk cost by the predicted cheese weight, and compare. A milk at 20% more per litre with 30% more yield is the cheaper cheese. This is a small calculation that quietly changes which supplier makes sense, and almost nobody does it.
A short buying checklist
- Not ultra-pasteurised. The single most common wasted batch.
- Whole rather than reduced-fat, unless the recipe specifically calls otherwise — the yield difference is large.
- Fresh. Milk close to its date has already begun acidifying; a titration will show it, and it changes how the make behaves.
- Note the fat and protein before you open it, so you can predict the yield and compare against what you actually get.
- Know the heat treatment and the source, and if you are giving the cheese to anyone else, be able to tell them.
Get the milk right and most recipes work. Get it wrong — particularly the ultra-pasteurised trap — and no amount of careful technique downstream will produce a curd.
How much milk do you actually need?
The most practical use of all this is working backwards from what you want. A kilo of hard cheese at 38% moisture from whole cow milk needs roughly ten kilos of milk — the calculator gives 1.01 kg from 10 kg. A kilo of a softer 50% cheese needs closer to eight.
That figure is worth having before you shop, because milk is bought in fixed containers and the difference between four litres and six is the difference between a wheel worth aging and an awkward small one that dries too fast.
Seasonal variation is real
Milk composition shifts through the year. Fat and protein are generally lower during flush, when animals are on fresh spring pasture and producing most heavily, and higher in late lactation. The swing is large enough to notice in a yield.
If you make the same cheese repeatedly and keep a note of the predicted and actual yields, that pattern becomes visible over a year — and it explains batches that came in light without any error in technique. It is also an argument for reading the carton each time rather than trusting a figure you memorised.
One more reason to check the carton
Beyond fat, protein and heat treatment, the date matters. Milk close to the end of its life has already begun acidifying, which a titration will show and which changes how the make behaves — the culture starts from a different baseline and the recipe's timings drift.
Fresh, whole, not ultra-pasteurised, from a source you know. Those four checks cover almost everything milk can do to a batch before you have added a single ingredient.