What Carryover Cooking Really Means
The dish continues cooking after the heat is off. Most cooks know this. Most still pull at the wrong moment.
A roast pulled from the oven at 55 degrees Celsius does not stay at 55 degrees. It climbs, often to 62 or 63, while it rests on the carving board. The pan that came off the burner is not finished cooking the steak inside it. The salmon fillet set down on the warm plate continues to firm under its own heat for another thirty seconds. Most cooks know this in some general way. Most still misjudge the moment of removal by the amount of the climb, and they overshoot every roast and every fillet they cook for the rest of their lives without ever realising the failure has a name.
The name is carryover cooking. The mechanism is straightforward thermodynamics. While food is on the heat, the outer layers are hotter than the inner layers — a temperature gradient runs from the surface inward. When the heat source is removed, the gradient does not vanish. The hotter outer mass continues to conduct heat into the cooler center until the food reaches thermal equilibrium with itself. The internal temperature at the geometric center, the place where the cook pushed the thermometer, will keep rising for minutes after the food leaves the oven. For a thick cut, the rise is five to ten degrees. For a thin one, the rise is a degree or two. For a very thick one — a standing rib roast, a leg of lamb — the rise can be a full ten to fifteen degrees and continues for the better part of an hour.
Two terms worth defining. Thermal mass is the capacity of a material to store heat, roughly proportional to its weight and specific heat capacity; a thick roast has more thermal mass than a thin fillet, which is why it carries over more aggressively and for longer. Equilibration is the process by which a temperature gradient inside a single object resolves itself toward uniformity once the external heat source is removed. Carryover cooking is, in physical terms, the equilibration phase of a roast.
Three variables govern how much carryover any specific dish will see. The first is thickness. A two-kilogram roast carries over far more than a thin fillet of fish, because the gradient has more material to move through and more reservoir to draw from. The second is the temperature of the outer surface at the moment of removal. A roast pulled from a 230-degree oven has a hotter crust than the same roast pulled from a 110-degree oven, and the hotter crust drives a steeper internal climb. The third is how the food is rested. A roast covered with foil retains its surface heat and pushes more energy inward; a roast left open dissipates surface heat to the air and carries over less. None of these variables is exotic. All three are visible to the cook, if the cook is looking.
The practical translation is a small piece of arithmetic. For a beef roast targeted at 55 degrees internal — rare in any reasonable definition — pull at 47 to 50 degrees, rest for five to ten minutes loosely covered, and let the temperature climb to 55 on the board. For a salmon fillet targeted at 55, pull at about 50 to 52; the climb is brief because the fillet is thin and its thermal mass is small. For a whole chicken or turkey where safety matters — the USDA target for poultry is 73 degrees in the breast — pull at 71 to 72 and let the rest finish the job. The cook does not need to memorise every number. The cook needs to memorise the principle: target the temperature you actually want when you EAT, not when you pull from heat. Subtract five to ten degrees, every time, depending on thickness.
The home-cooking translation of this physics has been worked out most carefully in the American test-kitchen tradition. Cook's Illustrated and the related output of America's Test Kitchen have spent decades publishing carryover numbers by cut, by oven temperature, and by resting protocol, and their guidance is among the most reliable available to the English-language home cook. The physics underneath — the equation that governs how heat moves through a slab of meat by conduction — is the same Fourier heat equation that any first-year engineering student meets, and Hervé This has written about its culinary implications at length, most accessibly in Building a Meal. Both sources, the practical and the theoretical, point at the same conclusion: a cook who ignores carryover is cooking blind in the final minutes of every dish.
There is a difficulty in the kitchen that the textbook does not name, which is that carryover interacts with another effect cooks already know about — see Why Temperature Is the Hidden Variable in Cooking for the broader argument about why the thermometer beats the clock. The interaction matters because the same probe that gives the cook precision while the meat is on the heat must also be trusted to estimate the climb during the rest. The estimate is not difficult once the cook has done it a few times, but it does require the willingness to pull "early" on the thermometer, which feels wrong to anyone trained by recipes that report a finishing temperature rather than a pulling temperature. The cookbook's number is the dinner-plate number. The pulling number, the one that actually matters, is five to ten degrees lower.
There are several views on this. Some chefs pull early — sometimes very early — and let carryover do the final work, on the theory that the gentler arrival at temperature produces a more even doneness from crust to center. Other chefs pull at the target and accept the overshoot as a small price for not having to estimate the climb. Still others, particularly in the French restaurant tradition, finish with a brief return to high heat after a long rest, to recrisp the surface and re-thermalise the interior on the cook's terms rather than physics's. My view is that the question is not which school is right, but which target the cook is aiming at. Target the temperature you want at the moment of eating. Subtract five to ten degrees for carryover. Pull. Rest. Eat at the target. The argument about styles is downstream of getting that single subtraction right, and the same principle underlies the broader practice described in The Science of Resting Meat — that the rest is not a pause between cooking and serving, but a continuation of cooking by other means.
The roast is still cooking on the board. The cook who knows this pulls earlier than instinct says, and dinner improves the moment the habit takes hold. The thermometer is not the only instrument that matters here. The clock matters, too — not the cooking clock, but the resting clock — and the cook who reads both is no longer guessing.