Alcohol in Cooking: What It Actually Does

Alcohol is a solvent for aroma

That is the first reason to use it, and the reason substitutes are hard.

Many aroma compounds dissolve poorly in water and poorly in fat. Ethanol sits between the two, dissolves both, and evaporates faster than water — carrying the aroma it dissolved into the air with it. A sauce made with wine smells wider than the same sauce without, from identical ingredients.

If blooming spices moves aroma into fat, alcohol is the tool that reaches the aromas stranded between fat and water.

It does not all cook off

This is the common misconception. Flaming it or simmering it for a long time still leaves a good deal behind. Measured: about 40% remains after fifteen minutes of simmering, and 5-10% after more than two hours.

For flavour that is negligible. For food a child is eating, or someone avoiding alcohol, it is the whole answer — and the only way to remove it completely is not to add it.

Setting it alight counts here too. Greek Saganaki is sometimes flamed with brandy at the table — a Chicago invention of 1968 rather than a Greek one — and the flame going out does not mean the alcohol has gone. What burns is the vapour at the surface; what is in the liquid stays there. Fine as theatre, wrong as a method for removing alcohol.

How it removes "off" smells

The fishy and gamey notes in seafood and meat come largely from volatile amines like trimethylamine, and those evaporate along with the alcohol.

So when that is the goal, add it early and leave the lid off. Covered, it condenses and comes straight back. Adding alcohol and immediately clapping the lid on does the opposite of what you wanted.

That is the sake in a karaage marinade and the soju in the pot when jokbal is simmered. Soju is useful here precisely because it has almost no aroma of its own — it removes without adding.

That pot has a second agent in it doing the opposite thing. Alongside the soju, jokbal puts in a tablespoon of doenjang — and a paste cannot evaporate, so nothing leaves with it and the lid is beside the point. It works by bringing a savour heavy enough to sit over what is underneath, which is why dak-hanmari, using neither, says it has "nothing to cover anything up". Removing and covering are different jobs, and one pot often wants both (fermented pastes).

Each kind does a different job

• Mirin — a seasoning wine, over 40% sugar, with alcohol. You are buying sweetness and gloss: the sugar reduces into a coating on the surface. That shine on teriyaki chicken and gyudon is mirin.
• Sake — almost no sugar; aroma and deodorising. Swap sake in where a recipe wants mirin and you lose the glaze.

That same sugar also decides when mirin goes in. Nikujaga adds mirin and sugar ahead of the soy, because sucrose is a far larger molecule than the ions salt breaks into and takes longer to travel inward (what sugar does). Where the gloss is the goal it goes the other way and arrives last — the tare on Jangeo-gui is layered on in repeated coats as it grills.

• Shaoxing — a Chinese yellow wine with a strong fermented aroma. It is the note underneath hong shao rou and mapo tofu; replace it with sake and the dish gets thinner.
• White wine — acidity and fruit. Seafood and cream. Bouillabaisse and risotto. Where the ingredient supplies its own liquid, though, the quantity collapses: only one glass goes into Moules-Frites, because the mussels pour out the seawater held in their shells. More wine there only dilutes that.
• Red wine — tannin and colour. Tannin needs long reduction to soften, so it does not belong in a quick dish. It suits long cooking like beef stroganoff.
• Beer — bitterness and fermented aroma, and in batter the carbonation helps it puff. Carbonade flamande uses it as liquid; fish and chips uses it as batter.
• Makgeolli — a category of its own. See below.

When the alcohol is there for the yeast, not the ethanol

Everything above treats drink as a liquid ingredient. An unfiltered drink also contains live yeast, and some dishes are built on that.

Jeungpyeon raises a rice batter with makgeolli. No baker's yeast is added: the yeast already in the drink making carbon dioxide over four or five hours is the only leavening there is. Which means pasteurised makgeolli will not raise it at all — half of all failures with this dish are exactly that.

Where the point is aroma or deodorising, one bottle substitutes for another. Where the point is the culture, nothing substitutes. The thing to read on the label is not the strength but the word "pasteurised" (fermentation basics).

When you add it decides what it does

The same bottle gives you three different results.

• Early, uncovered — deodorising. Driving the alcohol off is the point.
• Midway — the aroma stays. The sauce gains depth.
• At the end — the raw bite of alcohol stays too. Usually a mistake.

The commonest error is a hurried splash at the finish. It needs at least a minute or two of simmering to lose its edge.

A fourth moment — before the heat goes on at all

All three of those are about the pan. Alcohol also goes in before any cooking starts: German Sauerbraten sits beef in vinegar and red wine for three to four days.

There the alcohol is not something to drive off. It works as a solvent, moving aroma across over days — and that liquid then becomes the sauce. A recipe that throws the marinade away is not this dish.

So there are really four moments: before cooking (to move aroma across), early (to drive it off), midway (to keep it), and at the end (usually a mistake). Boiling a used marinade before serving it is partly hygiene and partly the same one or two minutes that takes the raw edge off the alcohol (what marinades do).

Where the job is to keep something from growing, strength is the whole of it

Every job on this page so far is measured by what the alcohol carries in or by what it leaves behind, and neither question is about how strong it was: whatever goes into a pan either evaporates or stays as a trace, and the percentage on the bottle never enters the reasoning. There is one more job where it is the only number there is, and no heat is involved at any stage of it.

• The alcohol is the medium, and its strength is a quantity you can lose. Uroe-jangajji buries salted melon in sake lees for months, and it cites this page while saying so: the four moments above "eventually meet heat. Nothing here is heated at any stage", so the eight percent or so left in the lees "stays at full strength and preserves the melon alongside the salt." The caution two sections up — that it does not all cook off — is the design here. And the number falls on its own: the water coming out of the melon "drops the alcohol and the sugar concentration together", so the used bed is thrown out and replaced rather than topped up, because stirring fresh lees into a thinned one "only averages the two". It is the same reason the recipe says never to loosen stiff lees with water — "that one cup drops the alcohol concentration far enough to open the door to mould." Loosen it with soju.
• Or it goes onto the surface and leaving is the point. Umeboshi wipes each plum with a dry cloth and then "again with a cloth dampened in shochu. Water left on them is where mould starts." Gotgam sprays the hanging persimmons with soju or vodka every morning for the first three days, because "While the surface is still wet is when mould takes hold most easily." In neither does the spirit become an ingredient in the dish.

That second one is the property this page opens with — ethanol evaporates faster than water — used with nothing dissolved in it. What it carries off is the water, and it leaves nothing behind to taste. So the two facts at the top of this page are those same facts read as virtues, and which one a recipe wants is settled by whether the alcohol is the medium or a swab: lees at eight percent has to hold its concentration against the food's own water for months, while a spirit at drinking strength on a cloth does its work on contact and is gone shortly after.

Salt does this job too and is better documented for it — fermentation basics puts umeboshi's 20% at the concentration where "neither lactic bacteria nor mould grows", and salt and seasoning has what the same salt does at fifteen times seasoning strength. And some dishes run both at once in different places: umeboshi takes 20% salt in the crock and the spirit on the surface before the fruit goes in, uroe-jangajji 25% salt in its first stage and eight percent alcohol in the second.

Substitution ratios that actually work

• 1 tbsp mirin ≈ 1 tbsp sake + ½ tsp sugar. Less gloss, closest match.
• 1 tbsp mirin ≈ 1 tbsp water + ½ tsp sugar + a few drops of vinegar. The alcohol-free version: it imitates the sweetness and acidity and none of the aroma.
• Sake for shaoxing works, but you have made a different dish.
• Water and vinegar for wine is not a substitution — you get acidity with no aroma. Use stock and finish with a few drops of vinegar instead (balancing flavour).

Korean "cooking wine" usually contains salt

Many supermarket cooking wines are seasoning products with salt and sugar added. They are not a 1:1 swap for mirin, and you have to taste before salting.

If the label lists sodium, it is a seasoning product and the dish needs less salt than the recipe says (salt and seasoning). This is one of the more common reasons a recipe followed exactly still comes out salty.