Casein vs Whey in Gelato: Which Protein Adds Body?
Table of contents
Every litre of milk you weigh into a base carries two protein systems that behave nothing alike. One survives the pasteuriser untouched and holds the emulsion together. The other unfolds in the vat and goes to work on the air cells. Knowing which is which changes how you build body.
Same source, two jobs. Most recipes only ever think about the total.
What the two proteins actually are
Whole milk carries roughly 3.3 g of protein per 100 g, and that protein splits about 80 percent casein to 20 percent whey. Marshall, Goff and Hartel give the same ratio in Ice Cream, and USDA FoodData Central puts the total protein figure in the same place.
Casein is not a single molecule floating in solution. It travels as micelles: aggregates of alpha-s1, alpha-s2, beta and kappa caseins held together by colloidal calcium phosphate, with kappa-casein forming a hairy outer layer that keeps the micelles from clumping. They are colloidally dispersed rather than dissolved, which is why milk is opaque.
Whey proteins are the ones left in solution after casein is removed. Beta-lactoglobulin is roughly half of that fraction, alpha-lactalbumin around a fifth, with serum albumin and immunoglobulins making up the rest. Unlike casein, these are compact globular proteins with an ordered structure, and that structure is the thing that heat destroys.
Where each one works: the fat surface and the air cell
Protein in a frozen dessert has two interfaces to defend. The first is the fat globule surface created during homogenisation, where freshly split globules need a membrane before they can re-coalesce. The second is the air cell wall created in the batch freezer.
Casein does most of the work at the fat surface. Micelles adsorb quickly and build a thick, mechanically robust membrane. Whey proteins adsorb there too, but casein wins the competition for surface area in a normal dairy base, partly because there is four times more of it and partly because a micelle is large and flexible enough to anchor across a wide patch of the globule.
The scale of that job is easy to underestimate. Homogenising a base cuts the average globule down towards a micron, and every halving of diameter roughly doubles the total surface area that needs covering. A recipe that lands 7 percent fat creates a great deal of new interface in a few seconds, and the protein has to be there when it happens. Under-supplied membranes are why some lean, low-solids bases churn to a greasy rather than a creamy texture.
At the air interface the ranking flips. Whey proteins, once they have unfolded, are the better foamers: they spread at the air and water boundary and form a viscoelastic film that keeps the air cells from collapsing. That is a large part of why a base with a healthy MSNF level holds overrun better than a lean one.
Quick reference. Milk protein is roughly 80 percent casein and 20 percent whey. Casein dominates the fat globule membrane and survives pasteurisation intact. Whey proteins denature from about 65 to 70 degrees Celsius upward and are the stronger stabiliser of air cells.
Figure 1. Same milk, two proteins, two different interfaces to defend.
Heat changes only one of them
Casein micelles are remarkably heat stable. Standard pasteurisation leaves them essentially intact, which is why you can hold a base at 85 degrees without the casein falling out. What does bring casein down is acid: at its isoelectric point near pH 4.6 the micelles lose their charge and precipitate, which is exactly what happens when you push too much fruit acid into a dairy base.
Whey proteins are the opposite. Beta-lactoglobulin begins unfolding in the 65 to 70 degree region and denatures extensively above 75 to 80 degrees. The unfolded protein exposes a free thiol group and binds to kappa-casein on the micelle surface. That reaction is not damage. It is the reason a base pasteurised at 85 degrees for a short hold feels fuller than the same recipe pasteurised at 65: you have converted soluble whey protein into a water-binding complex sitting on the casein micelle.
This is the single most underused lever in a small lab. The pasteurisation curve is a texture control, not just a food safety step. A high-temperature short-time hold denatures a larger share of the beta-lactoglobulin than a low-temperature long hold at the same lethality, and you can feel the difference in the finished scoop. If a base tastes correct but drinks thin off the spoon before churning, the pasteurisation profile is worth checking before you reach for more powder.
The corollary matters too. Once whey protein has been denatured in the vat it cannot be denatured again, so a base built from reconstituted powder that was already heat treated during drying starts the day with less functional whey than a base built from fresh milk. Powder grades are classified by exactly that: low-heat, medium-heat and high-heat skim milk powder differ in how much undenatured whey protein nitrogen survives, and the number is on the specification sheet if you ask for it.
What that means for body, and what it costs
More protein sounds like an obvious win. It is not free.
The first ceiling is lactose. MSNF is roughly 54 percent lactose, so raising milk solids to chase protein raises lactose in lock step, and lactose is the least soluble sugar in the mix. Push it too far and it crystallises during storage into the gritty defect described in sandy gelato texture. This is why milk protein concentrate and whey protein concentrate exist as separate ingredients from skim milk powder: they let you add protein without dragging lactose along.
The second ceiling is the fat surface itself. Protein that is too well anchored on the globule resists being displaced, and displacement is what emulsifiers are for. Over-stabilise the membrane and partial coalescence never happens properly: the gelato comes out wet, slumps in the pan and reads thin despite a respectable fat figure.
The third is flavour. Whey protein above a modest dose brings a cooked, faintly chalky note that no amount of vanilla covers. Anyone who has built a high protein gelato has met it.
Dosing them in a real base
| Ingredient | Protein content | Lactose load | What it is good at |
|---|---|---|---|
| Skim milk powder | about 34 percent | high, about 52 percent | cheap MSNF, general body |
| Milk protein concentrate | 56 to 85 percent | low | casein-led body, no lactose penalty |
| Whey protein concentrate | 34 to 80 percent | varies by grade | foam stability, overrun |
| Sodium caseinate | about 90 percent | negligible | emulsion strength |
| Egg yolk | about 16 percent | none | emulsification plus flavour |
A workable starting point in a white base is total MSNF at 9 to 11 percent from milk and skim milk powder, then a small targeted addition rather than more powder if the body is still short. One to two percent of a milk protein concentrate lifts casein without touching lactose. A similar dose of whey concentrate lifts overrun stability instead. Doing both at once usually overshoots.
When to reach for which
Reach for casein-led protein when the problem is a thin, fast-melting scoop with adequate fat: you need a stronger fat globule membrane and more bound water. Reach for whey when the problem is air that will not stay in, a scoop that deflates in the display case, or a lean base that needs foam structure it cannot get from fat.
And if the base is already at the lactose ceiling, neither answer is more powder. It is a concentrate, or it is condensed milk versus milk powder arithmetic done properly.
Related Concepts
- What is MSNF sets the frame both proteins live inside
- Why is my gelato gummy covers what happens when protein and stabiliser both go too far
- Pea protein in vegan gelato is the plant-side version of the same decision
- Lactose-free gelato shows what changes when the lactose ceiling moves
