What Is In It

This is dessert. It's a processed food, and I won't call a jar of milk and sugar healthy. Nobody should.

It is also one of the simplest processed foods you can buy: milk and sugar cooked for hours, plus a short list of helpers. This year, during an FDA visit to our facility, the ask came through: update the labels. The updated list names every ingredient in full legal language — and legal language is not kind. "Interesterified soybean oil." "Distilled monoglycerides." I read the new list and I didn't like looking at it.

So before printing it, I stopped and had every single ingredient researched properly. Regulatory opinions from the FDA, the European Food Safety Authority and the World Health Organization first, then independent studies with public funding. One rule ran through all of it: a study paid for by the people who sell the ingredient doesn't count in its favor.

Here is what came back. All of it, including the parts that don't help me.

The science isn't finished on everything in here, and where it isn't, I say so.

---

The label, as it prints

The legal ingredient statement, unedited.

The first four lines are the food: sugar, water, milk and a fat. Everything after them — every name that looks alarming — is the small stuff, and I've printed how small it is on the pail itself.

How little of it there is

Ingredient by ingredient. The pail carries the total; this is what's behind it.

Ingredient	How much	What it does
Sodium bicarbonate	< 0.1%	baking soda; controls acidity in the cook
Agar-agar	< 0.1%	a seaweed; body and texture
Potassium sorbate	< 0.1%	stops yeast and mold
Distilled monoglycerides	< 0.1%	a fat; keeps the milk from boiling over
Lactase	< 0.02%	an enzyme; keeps it from turning gritty
Natural vanilla flavor	< 0.02%	the vanilla
Lactose	trace	milk sugar
All of them together	< 0.5%	less than an ounce in this eleven-pound pail

---

The list, in plain words

Every ingredient on the label, in the order it prints.

Sugar

First ingredient by weight, and the only thing on this list that really matters to your health. I'll get to it.

Water

Tap water. The same water the city drinks — nothing fancier than that.

Nonfat dry milk

Milk with the fat taken out and the water dried off. My abuela made dulce de leche by boiling down fresh whole milk in an open pot for hours. Mine starts from milk powder and water instead: powder doesn't sour on its way to the kettle, doesn't change with the season, and makes every batch come out the same. Add the water back — that's milk.

Interesterified soybean oil

Soybean oil rearranged so it stays solid at room temperature without being hydrogenated. It is in there because the milk I start from has no fat left in it, so the fat gets added back — as a solid vegetable fat that holds up the way a bakery needs it to: through an oven, and through weeks of an open pail at room temperature. Because it comes from soy, the label says CONTAINS SOY. That is the FDA's rule and I follow it to the letter. This is the one I have the most to say about, further down.

Sodium bicarbonate · less than 0.1%

"Sodium bicarbonate" may be the scariest-sounding name on this list. It's baking soda — a mined mineral, the same one in your kitchen. It controls the acidity of the milk while it cooks; that is its whole job in the kettle. And it is the one ingredient on this list that was also in my abuela's pot.

Agar-agar · less than 0.1%

Seaweed — red algae, cooked in hot water, filtered, and dried. Some agar gets an alkaline wash before the cooking; no solvents are involved either way. As industrial ingredients go, this is about as close to "just a plant" as it gets, and it is permitted in USDA certified-organic food. It gives the dulce de leche its body and keeps it from weeping liquid on the shelf.

Potassium sorbate · less than 0.1%

A preservative — first found in rowan berries. It stops yeast and mold. Your body treats it like a fatty acid: it burns most of it and breathes it out as carbon dioxide within a couple of hours.

Distilled monoglycerides · less than 0.1%

Chemically this is a fat — glycerol with fatty acids attached, which is what every fat in your kitchen is, just with more of them on each molecule. Your gut digests it with the same enzyme, into the same two pieces, as any other fat you eat. Its job in my product is simple and physical: the vat foams when the milk cooks, and this keeps it from boiling over. I have more to say about this one below.

Lactase · less than 0.02%

The enzyme in Lactaid, brewed from a dairy yeast. It splits milk sugar into two simpler sugars so the dulce de leche doesn't turn gritty after a few months on a shelf — and then the cooking destroys it. It is on the label because US law lists what goes into a food, not what survives the process: the same reason you see baking soda there when nearly all of it has reacted away by the time the batch leaves the kettle.

Natural vanilla flavor · less than 0.02%

Vanilla flavor carried on cane sugar and gum arabic. Straight up: most of what is doing the flavoring here is not vanilla bean. US law lists the parts of a compound ingredient heaviest first, and on mine "natural flavor" comes before "vanilla bean extractives." The bean is the smaller part. Real vanilla bean is the truffle of the food world — rare, expensive, and for most of us more myth than pantry. This is how nearly everyone actually gets to taste it. I'd rather tell you that than sell you the myth.

Lactose · trace

Milk sugar, dead last on the list, and a trace. Nobody added it — it came in with the milk, and the lactase just above splits part of it into simpler sugars. What's left is still there, so US law says it goes on the list. On the list it goes.

---

Why the list is this long

It is not eleven separate decisions. It is one job: taking a recipe that was made in an open pot and eaten the same week, and getting it into an eleven, twenty-two or fifty-five pound pail that a bakery opens and works from for weeks — the label says refrigerate after opening; the recipe has to survive the days when nobody does. Every helper on this list is there because it is the only way I know to do that to my abuela's recipe safely and consistently — same taste, same texture, every batch, for months. The agar supplies the body. The preservative exists because of the open pail. The monoglycerides are there for the boil-over. Take the pail and the weeks away, and most of the list goes with them.

And the pail itself is cooked for about three hours, the way dulce de leche has always been made at home. Hours at a boil leave bacteria very little chance. They leave it just as little chance of improving your diet. That is the plain truth about this processed food: careful, consistent, and dessert.

---

The two I can't fully clear

I'd rather tell you this than have you find it.

Interesterified soybean oil

No long-term human study has ever been run on this specific fat. Not one. The paper most often quoted against it, from 2007, was funded by the Malaysian Palm Oil Board. Its lead author was employed there, its second author was a graduate student there, its third author sat on the Malaysian Palm Oil Advisory Council — and the fat they compared it against was palm oil, which is the product the board exists to sell. The best independent work I could find is a 2020 systematic review that declared no funding from anyone and found no effect on blood cholesterol. Against that sit two studies from Brazilian public universities that found effects on the livers of rats. Rats aren't people, and a review that finds nothing isn't the same as proof of safety. The truthful word for this fat is unresolved. Not safe, not dangerous. Nobody has done the work.

Texas and Louisiana have both put interesterified soybean oil on a list of ingredients that will require a warning label or a QR code — Texas from January 2027 under Senate Bill 25, Louisiana from January 2028. That is a legislature's judgment, not a scientific finding, and four food-industry trade associations are fighting the Texas law in federal court right now — a judge has blocked enforcement against them while the state appeals. But it is on the list, and I am not going to pretend it isn't.

Distilled monoglycerides

A French study called NutriNet-Santé followed 92,000 adults, on French public money, and found that people who ate more of this class of emulsifier had modestly more cancer. My honest read: distilled monoglycerides are in nearly every packaged baked good on earth, which makes them a very good marker for somebody who eats a lot of packaged food — and the two emulsifiers with the strongest evidence of harm in animal studies showed nothing at all in that same French study, which is what you would expect if the association is the shadow of something else rather than the thing itself. That is my read, and I want to be clear that it is a read. I can't prove it. Neither can they.

---

The part nobody in my business says out loud

Every additive on that list, all of them together, comes to less than half of one percent of the jar — less than an ounce in an eleven-pound pail. The rest is sugar, milk and the fat. Dulce de leche is about half sugar by weight and always has been.

To reach the European safety limit for potassium sorbate — a limit that already has a hundred-fold margin built into it — you would have to eat 210 grams of this dulce de leche every day for the rest of your life. Those 210 grams carry more than twice the sugar the World Health Organization says a person should eat in a whole day.

The preservative would be the least of your problems. If you are going to worry about this jar, worry about the sugar.

Sugar is a horrible ingredient. At least I say so — and I make it taste better than anyone else does.

This is dessert. It has been dessert since 1988, and it will go on being dessert.

---

Every study, and who paid for it

The whole evidence base this page rests on. Funding is flagged in both directions — industry and advocacy.