Scope. This page provides a controlled reader-run protocol and does not claim site-run measurements. For cooking, cooling, storage and reheating, follow the current official guidance linked on the page.
Tempering is taught as a sequence of movements. Warm the milk. Whisk the yolks. Add a little hot milk to the yolks. Add a little more. Return the lot to the pan. Cooks who follow it get away with it most of the time and never learn what any of the steps were for, which means that the first time the ritual is inconvenient — one hand busy, a heavy pan, a doubled recipe — they cannot work out which parts they are allowed to skip.
Here is what tempering is doing, and what the controlling variable actually is.
What tempering is preventing, physically
Egg proteins coagulate when they are heated. That is the whole basis of custard, and it is also the whole problem: the same reaction that thickens the mixture will, given the chance, set into visible lumps of cooked egg suspended in warm liquid. Once a lump has set it stays set. There is no step later in the recipe that un-cooks it. Straining removes it and takes some of your custard with it.
So the failure mode is not "the mixture got too hot." The failure mode is "a small part of the mixture got too hot, briefly, before it could be diluted." Those are different problems with different solutions, and almost all confused tempering advice comes from treating them as the same one.
The differential that matters, and the one that does not
The number people think about is the bulk differential: hot liquid at one temperature, eggs at fridge temperature, and the gap between them. It feels like the risk lives in that gap. It mostly does not.
What decides whether you get lumps is the local differential and the contact time. When hot liquid meets egg, the first fraction of the egg it touches sees something close to the liquid's full temperature. It does not see the average of the two. If that thin layer of egg sits at that temperature for long enough to coagulate, it coagulates, and it does so regardless of what the mixture's average temperature works out to a second later.
This is why the two things cooks find counter-intuitive are both true:
- You can add very hot liquid to cold eggs safely if the eggs are moving fast and the stream is thin, because no parcel of egg stays in contact with undiluted hot liquid for long.
- You can curdle eggs with liquid that is barely warm enough to set them, if you tip it in as a slug and stir afterwards.
The risk is local, not average. Everything below follows from that sentence.
To put a number on it, measure the temperature at the contact point under three pour rates and two agitation levels, then record the bulk differential at which visible coagulation first appears with and without whisking. This page supplies no universal "safe differential"; measure it on your own setup, and distrust any number that does not say how it was obtained.
This article does not rely on isolated yolk, white or whole-egg set-point tables. The useful comparison is between the actual mixtures: a diluted custard behaves differently from scrambled egg, and the narrow window that decides a custard sets out how to record that difference on your own mixture.
The method, with the two things most people get backwards
The standard sequence, with the reasoning attached:
- Break the eggs down completely before any liquid arrives. Yolks whisked smooth, with the sugar already in if the recipe uses sugar. A yolk that is still partly intact is a parcel of concentrated protein waiting to set as a unit.
- Stabilise the bowl. A damp cloth ring under it, or a bowl that sits in a pan. You need both hands: one whisking, one pouring. This step is skipped more often than any other and it is the reason people end up pouring badly.
- Start the whisk before the liquid. The whisk should already be moving when the first drop lands.
- Pour a thin stream into the moving part of the bowl. Not down the side. Not onto the back of a spoon. Into the turbulence.
- Once the eggs are warm and loosened, the rest can go in faster. The margin widens as you go, because the mixture you are adding to is now closer to the liquid's temperature and much more dilute.
- Return everything to the pan and take the heat down. Tempering has solved the mixing problem. It has not solved the cooking problem, and the cooking problem is where most custards are actually lost.
The two things people reliably get backwards:
Backwards one: they pour, then whisk. Alternating pour-and-stir feels careful. It is the opposite of careful, because every pause is a period in which hot liquid is resting against still egg. Continuous agitation with a continuous stream is safer than intermittent everything.
Backwards two: they aim the stream at the side of the bowl to slow it down. The side of the bowl is the least agitated part of the bowl. Slowing the delivery and simultaneously delivering it into stationary egg trades one protection for the loss of a bigger one.
There is also a rule about direction that is worth stating plainly: hot goes into cold, not cold into hot. Adding eggs to a pan of hot liquid puts them in contact with hot metal as well as hot liquid, and the pan floor is the hottest surface in the operation.
Whisking rate versus pouring rate
These are the two control knobs, and they trade against each other:
| If you have | Then you can | Because |
|---|---|---|
| Vigorous, continuous whisking | Pour considerably faster | Incoming liquid is dispersed before any parcel of egg holds a high local temperature |
| One hand only, or an awkward bowl | Pour much more slowly | Low agitation means longer contact time at the interface |
| A second pair of hands | Effectively ignore the pour rate | Agitation and delivery are independently controlled |
| A blender or stick blender | Stream it in | Mechanical agitation far exceeds anything a hand whisk achieves |
The whisk is doing two separate jobs. It disperses incoming hot liquid, and it physically breaks up any egg that has begun to set before it becomes a lump you can see. The second job is the one nobody mentions, and it is why a mixture that gets a bit ahead of you can often be whisked back to smooth if you notice immediately.
The shortcut that works
Cool the liquid first. This is the shortcut, and it is almost never taught, presumably because it is not dramatic. Every bit of temperature you let the liquid lose before it meets the eggs reduces the local differential directly. If your infused milk has been sitting off the heat while you found the vanilla, that is not a delay, that is margin. There is no downside except time, and no technique required.
Two more that genuinely work:
- Use the machine. Eggs in the blender jug, motor running, hot liquid streamed in. The agitation is high enough that the pour rate largely stops mattering. The trade is aeration, which matters for some preparations and not others.
- Dilute the eggs before the liquid arrives. If the recipe contains cold cream, some of it can go into the yolks first. You have now made the egg mixture more dilute and cooler before the hot component appears, which widens the margin at both ends.
Sugar and starch change the formulation, but this article does not assign either one a universal shift in setting temperature. That comparison belongs with the custard window rather than in a tempering instruction.
The shortcut that does not work, and why it seems to until it doesn't
"Take the eggs out of the fridge an hour before."
Room-temperature eggs reduce the bulk differential by a modest amount. They do almost nothing to the local differential, which is the one that curdles custard. The reason this advice survives is that people who take the eggs out early also tend to be people who are cooking attentively — they whisked properly and poured reasonably, and they attribute the success to the step they consciously performed rather than to the two they did by instinct.
When it fails, it fails at exactly the point it would have failed anyway. And it fails worse than average, because the cook believed the protective step had already been taken and poured with confidence.
Room-temperature eggs are not a bad idea. They are just not tempering, and they should not be counted as tempering.
The same applies to pouring down the back of a spoon. It looks like a technique. It delivers hot liquid slowly to the part of the bowl the whisk is not reaching.
When you do not need to temper at all
Credibility here depends on saying where the technique is unnecessary, because a habit performed everywhere teaches you nothing about anywhere.
You can skip tempering when:
- The liquid is not hot enough to set egg. If you are combining something at or near room temperature, there is no reaction to prevent. Vinaigrettes, cold mousses and uncooked dressings do not need it.
- The eggs are already heavily diluted. A batter that is mostly liquid, or a mixture where the eggs are a small fraction of a large cold volume, has a built-in margin. Adding warm liquid to a full quantity of cold cream and eggs is not the same operation as adding it to four yolks.
- Everything rises together from cold. Some custards and egg-thickened sauces are built cold in one pan and brought up as a single mass. No differential ever exists. The variable there is rate of rise and the temperature you stop at, not mixing.
- You are using a machine that makes the pour rate irrelevant. Streaming into a running blender is not tempering carefully; it is making tempering unnecessary through brute agitation.
Tempering out of habit is harmless in itself. The cost is that it obscures which step was actually protecting the dish, and a cook who cannot name the protective step cannot adapt when the kitchen makes the ritual impossible. More of the ways this goes wrong are collected in nine things that kill a custard.
Frequently asked
Can I fix a mixture that has just started to look grainy? Sometimes, if you catch it in the first seconds. Whisk hard immediately, off the heat. Fine grain that has not yet become visible lumps can often be broken up. Once you can see distinct pieces of set egg, straining is the only remaining option and you will lose volume.
Does the shape of the whisk matter? A balloon whisk moves more of the bowl than a flat or coil whisk, which matters here because you are trying to eliminate still zones. This page makes no stronger whisk comparison; treat one as unmeasured unless it carries a controlled test.
Is a lumpy custard safe to eat? That is a food-safety question rather than a technique question, and we do not answer those from memory. The egg-safety guidance that named authorities publish is linked in raw-egg preparations.
In Plain English
What it means. Tempering manages the temperature where hot liquid touches egg, not the average temperature of the bowl. Lumps form locally and instantly, so the fix is local and instant too: keep the egg moving and the stream thin.
Who it helps. Anyone making crème anglaise, pastry cream, curd, egg-thickened soups and sauces, or anything where hot liquid and raw egg have to meet.
When to use it. Whenever liquid hot enough to set egg is going into egg. Not otherwise.
What to do next. Three changes cover most of it: whisk before you pour, aim into the moving part of the bowl, and let the hot liquid sit for a minute before you start. If you own a stick blender, use it and stop worrying about the pour rate entirely.
Next step
Tempering solves the mixing problem. It does not solve the cooking problem, which is the narrow window between a custard that coats a spoon and one that has curdled in the pan. That window is where most of them are actually lost.
Read the narrow window that decides a custard for how to find the working range and the failure point, and what a spoonful of starch changes. More procedural pages are collected in the technique library.
Related technique: Your Scrambled Eggs At Eight Temperatures.