Braising: Collagen, Time, and Why It Cannot Be Rushed

Tough cuts turn tender not because the muscle relaxes but because collagen becomes gelatin — and that conversion needs time, not more heat.

Two proteins moving in opposite directions

Meat contains two things that behave inversely under heat.

  • Muscle fibre contracts and squeezes water out. The longer you cook it, the drier it gets.
  • Collagen (connective tissue) breaks down into gelatin given heat and time. The longer you cook it, the more tender it gets — and the gelatin gives the liquid body.

A steak only has to account for the fibres, which is why it is cooked briefly. Cuts full of collagen — short rib, shin, collar — instead accept the drying of the fibres in exchange for converting collagen to gelatin, and that gelatin reads as succulence in the mouth.

So braising is not "cooking less"; it is cooking well past the point of no return. Stopping halfway is the worst outcome: the fibres have already tightened and the collagen has not yet let go.

Time matters more than temperature

Collagen converts slowly, above roughly 70°C. Pushing to 100°C does not speed it up much, and it contracts the fibres harder, so the meat gets drier.

Hence the standard: just below a simmer (85–95°C), for a long time. Boiled hard, you get the contradictory result of meat that falls apart and is still tough.

  • Keep the surface barely trembling.
  • In an oven, 150–160°C with the lid on.
  • A pressure cooker runs at ~120°C and cuts the time to a third — but it cannot brown, so sear first.

The sequence

1. Sear the surface. Aroma, and fond in the pan. Why in batches: Browning. 2. Cook the aromatics. Give the onion real time. 3. Deglaze. Wine, beer or stock, scraping the base. Carbonnade Flamande does this with beer. 4. Liquid only two-thirds up the meat. Fully submerged is boiling, and the flavour leaves for the liquid. The exposed top cooks in steam. 5. Lid on, low, two to three hours. 6. Judge it with a fork, not a thermometer — no resistance means done.

Some dishes break that sequence wholesale, and each for a different reason.

  • Skipping step 1 entirely. Blanquette de Veau does not sear the veal. It comes up from cold water, gets skimmed, and that water is thrown away before anything else — because any browning would make the sauce not white. Fårikål skips 1, 2 and 3 together and just layers lamb, cabbage, salt and whole peppercorns; the cabbage stands in for the aromatics.
  • Breaking step 4 on purpose. Blanquette submerges the meat completely. "The flavour leaves for the liquid" is the objective there, because that liquid becomes the sauce. Which of the two you are keeping decides which way the rule points.
  • Why step 4 actually matters when it does. The liquid in Dongpo Rou comes only two-thirds up the belly because the skin has to cook in steam above it. That skin is nearly pure collagen, so submerged its gelatin leaves for the pot and what is left is slack rind.
  • Step 5 is the one step this list never explains. "Lid on" is doing a job, and the job is holding step 4 where you put it: what evaporates has to come back, or the line falls over three hours and the exposed top stops cooking in steam and starts drying instead. A lid can be built for that. Tagine seals nothing — the tall cone simply returns the condensate to the rim rather than onto the meat, so the liquid barely drops across an hour and the pot is never stirred (steaming has the general case, and dry heat and moist heat the sealed family). The reverse is a tool too: a tagine's last ten minutes go uncovered, which is step 5 switched off on purpose once reduction is what you want.
  • Where step 6 does not apply. The octopus in Polbo á Feira has neither a target temperature nor a fork test. Rubber if under, mush if over, with a narrow band between — only a skewer tells you. And what tenderises it is not heat but freezing (freezing and thawing).
  • Where not stirring is the rule. Cassoulet so the beans stay whole, Fårikål so the cabbage wedges survive, Ajiaco so three kinds of potato do not become one purée — three dishes, three different reasons.

And the whole sequence assumes the thing you are softening is collagen

Steps 2 and 3 both put acid into the pot early — tomato with the aromatics, wine to deglaze — and step 6 says that if it is still tough it needs longer. For collagen both are safe, because time is always on your side. For an ingredient held together by pectin instead, neither is. Acid sets pectin firm, so the endpoint stops being a matter of time and moves out of reach altogether: the pot can simmer all afternoon and never arrive (Soaking has the chemistry).

Where a dish here braises a collagen-free ingredient for the same hours, it holds the acid back.

  • Red Red simmers its beans "in fresh water for 40 minutes with no tomato in the pot — acid sets pectin firm and the beans will never soften".
  • Cassoulet, cited just above for not stirring, says it in its own first step: "nothing acidic goes into this pot".
  • Abgoosht is the sharpest case, because both mechanisms sit in one pot: lamb shank whose collagen wants the hours, and chickpeas whose pectin wants the tomato kept out. So the order splits in two — two hours with neither salt nor tomato, then the tomato, dried limes and potato for the last forty minutes.

So before reaching for more time, ask what is actually holding the thing together. The cut list at the end of this guide answers that question for meat only.

Success is liquid that clings to your lips

The sign that enough gelatin has come out is broth that feels slightly tacky on the lips, and sets like jelly when chilled. That is what the liquid in Irish Stew or Galbijjim does.

If it is still watery after hours, either the cut had little collagen or it needs longer.

Why it is better the next day

  • Gelatin and fat set, which makes skimming the fat off easy.
  • Aroma compounds redistribute through the fat and liquid.
  • The meat reabsorbs the braising liquid.

Goulash, Rendang and Carbonnade Flamande are all understood as next-day dishes. Making them the day before is the standard approach, not a compromise.

And this is not only about braises. It holds across anything with a liquid, and collagen is not the only mechanism — a porous solid contracting as it cools and drawing the broth in is the more direct one. See Oden and Reviving leftovers.

Braises that break the pattern

  • Rendang: reduced until the liquid is gone. Only the coconut fat remains, so at the end the meat is effectively frying — a braise that crosses into dry heat.
  • Suyuk: boiling rather than braising, but with the same aim. Fully submerged, so the liquid is discarded, while the collagen softens. Slice against the grain — see Cooking meat.
  • Stegt Flæsk: it has to soften the collagen in pork belly and crisp the rind — both families on one plate.
  • Carnitas: hours submerged in lard rather than water. Water stops at 100°C and pulls the meat's soluble flavour into itself; fat does neither — which is why Suyuk ends up with a good broth and this ends up with the flavour still in the meat. Same pork, same hours, opposite results.
  • Eisbein: solves the leaching problem by moving the seasoning earlier, brining the knuckle for three to five days before it ever meets water.

Choosing the cut

Good for braising — beef: short rib, shin, chuck. Pork: collar, shoulder, belly. Chicken: thigh.

Bad for braising — fillet, sirloin, chicken breast. Little collagen, so long cooking just dries them. The expensive cuts are the wrong ones here.

The list says whether there is collagen, not how long it will take

Every entry above is a place on an animal, and the question it answers is whether there is connective tissue there. What it cannot show is the second variable — how far that collagen has already cross-linked, which is set by how old the animal was. The same cut off two animals is two schedules, and nothing on the list tells them apart.

One cut, three animals. Ossobuco is veal shank in two hours, and its own page prices the substitutes: "beef shin works but its connective tissue is tougher, so it wants three hours or more", while "pork shank is far cheaper and two hours is enough, but there is less gelatin and the sauce clings less." A swap is therefore not a matter of resetting the clock — the amount of gelatin at the end moves with it, and so does the body of the sauce.

One animal, two ages. Fårikål says mutton in place of lamb is "much stronger, and about thirty minutes longer — it was the original", and Rogan Josh that goat "takes about 30 minutes longer and gives a deeper sauce". Nihari runs it the other way: chicken thighs cut four or five hours "to an hour, though you lose much of that clinging body."

And a young animal can be chosen because the pot has no more time to give. Samgyetang cites this guide while saying so: a young bird's "collagen has not yet cross-linked into the tough form, so forty minutes is enough to turn it to gelatine. An older bird needs hours to reach the same place, and by then the rice has long since collapsed." The bird is stuffed with glutinous rice, so there are two clocks in one pot and the shorter one governs. Time being always on your side is true of the collagen and not of everything sharing the pot with it — so the age is picked to fit the schedule rather than the schedule stretched to fit the animal.

Which makes "good for braising" a range rather than a property. Chicken thigh is on the list above and rogan josh still turns the bird down: "the dish assumes bone and collagen that can survive 90 minutes of braising." Nihari's chicken version is the same judgement reaching the other answer — keep the bird, shorten the braise to an hour, accept the thinner sauce. Before asking whether a cut has collagen, ask how long this recipe needs it to last.

The same stiffening with age happens to the other polymer, and Soaking records it for beans: the ones that never soften are usually old, because "in storage the pectin in the skins cross-links permanently, and nothing reverses it." And where Stocks and broth gives a clock per ingredient, this is the case where one ingredient carries two.