What a temperature reading actually tells you
Two separate things happen as meat heats, and doneness language mashes them together. The first is texture: proteins tighten and squeeze out water, collagen begins converting to gelatin above roughly 160 °F given enough time, and fat renders. That is why a brisket held for hours at 200 °F is tender while a steak taken there is dry — same protein, different job. The second is pathogen reduction, which is a function of temperature and time held at that temperature. The published single numbers are instantaneous ones, chosen to be simple enough to use and conservative enough to be safe without a stopwatch.
Both things are invisible from the outside. Juice color, firmness by touch, and minutes per pound are proxies that fail in ordinary circumstances — a thicker cut, a colder starting temperature, a crowded oven, a lid opened three times. A decent instant-read thermometer removes the guessing for both the safety question and the texture question at the same time, and it is the least expensive upgrade available to a home cook.
The USDA-recommended safe minimums
These are the figures published by USDA Food Safety and Inspection Service as safe minimum internal temperatures. They are not my numbers and they are not stylistic preferences; they are the public-health baseline, and everything else on this page is measured against them.
| Food | USDA safe minimum internal temperature | Rest |
|---|---|---|
| All poultry — chicken, turkey, duck, whole birds, parts, ground, and stuffing cooked inside a bird | 165 °F (74 °C) | None required |
| Ground beef, pork, lamb, veal | 160 °F (71 °C) | None required |
| Beef, pork, lamb, veal — steaks, chops, roasts (whole muscle) | 145 °F (63 °C) | 3 minutes before carving or eating |
| Fish and shellfish with fins | 145 °F (63 °C) | None required |
| Ham, fresh or smoked, uncooked | 145 °F (63 °C) | 3 minutes |
| Egg dishes, casseroles, leftovers, anything reheated | 165 °F (74 °C) | None required |
Note the shape of that table: poultry is one number for every form, ground meats are one number, and whole cuts are lower. That is not arbitrary, and the reason is the single most useful concept here.
Why ground meat and poultry get different numbers
On an intact steak or chop, contamination is a surface phenomenon. Bacteria live on the outside of the muscle where knives, hands, hides and air touched it; the interior of an intact muscle is essentially sterile. Searing the outside of a steak takes that surface far past any of the numbers above within seconds, which is why a whole cut can be served with a cool center under conditions where ground meat cannot.
Grinding destroys that geometry. It takes the surface of a large amount of meat and distributes it evenly throughout the mass, so there is no longer an interior that was never exposed. Every part of a burger is, in effect, former surface. That is the entire reason ground beef carries a 160 °F recommendation while a steak from the same animal carries 145 °F. The same logic applies to any cut that has been mechanically tenderized, needled, or brine-injected, because those processes also drive surface organisms inward — and that treatment is not always obvious from the package.
Poultry is treated as a whole category at 165 °F because the relevant organisms are commonly distributed through the bird rather than confined to the surface, and because the consequences of getting it wrong are severe. There is no home technique that makes an undercooked chicken thigh a controlled risk in the way a seared steak is.
Steak doneness is a preference, and it carries risk
Most people do not cook a steak to 145 °F, and it is worth being straightforward about what that means rather than pretending otherwise. The table below is a culinary reference. Every temperature in it below 145 °F is a personal preference that carries a real, non-zero risk of foodborne illness, and no home cooking method eliminates that risk. It applies only to intact, whole-muscle cuts, and never to ground meat, to mechanically tenderized or injected cuts, or to poultry.
| Culinary term | Pull from heat | After carryover and rest | Status |
|---|---|---|---|
| Rare | 120–125 °F | 125–130 °F | Well below the USDA figure. Preference, with risk. |
| Medium rare | 130–135 °F | 135–140 °F | Below the USDA figure. Preference, with risk. |
| Medium | 135–140 °F | 140–145 °F | At or near the USDA figure once rested |
| Medium well | 145–150 °F | 150–155 °F | Above the USDA figure |
| Well done | 155 °F and up | 160 °F and up | Above the USDA figure |
If you choose a doneness below 145 °F, the things that reduce the risk are: buying whole cuts rather than pre-cut or tenderized ones, avoiding anything labelled as needled, blade-tenderized or enhanced, searing every exterior surface hard including the edges, and cutting a roast from a single intact piece rather than from something reassembled. None of that makes it equivalent to a fully cooked piece of meat.
Who should not make that trade
Pregnant people, children, adults over about 65, anyone with a weakened immune system, anyone on chemotherapy or immunosuppressive medication, and anyone with liver disease or reduced stomach acid should follow the conservative guidance in the USDA table and skip the preference table entirely. For these groups the consequences of an infection that a healthy adult would shrug off are meaningfully worse, and the correct advice is the published minimum, not a negotiated version of it. That extends to runny eggs, raw fish, raw sprouts and unpasteurized dairy and juice as well. If you are cooking for a mixed table, cook to the safe minimum and let anyone who wants theirs rarer take the risk on their own plate rather than deciding for them — hosting a dinner covers the surrounding logistics.
Carryover cooking and resting
Meat keeps cooking after it leaves the heat. The exterior is far hotter than the center, and that heat continues to move inward until the gradient evens out. How much it rises depends almost entirely on mass and on how aggressive the cooking heat was.
| Item | Typical carryover rise | Rest time |
|---|---|---|
| Large roast — prime rib, pork shoulder, whole turkey | 5–10 °F, occasionally more | 20–40 minutes, tented loosely |
| Thick steak or chop, 1.5 inches | 3–5 °F | 5–10 minutes |
| Thin steak, cutlet, boneless chicken breast | 1–3 °F | 3–5 minutes |
| Anything cooked low and slow in a gentle oven | Minimal, because the gradient is shallow | Rest for moisture, not for temperature |
The practical consequence is that a large roast has to come out of the oven meaningfully below its target, and pulling one at the finished number guarantees overshooting it. The second consequence is a good reason to rest that has nothing to do with temperature: heat drives moisture toward the cooler center and tightens the proteins, and a few minutes off the heat lets both relax. Cut a steak the instant it leaves the pan and a puddle ends up on the board; give it five minutes and most of that stays inside.
Tent loosely with foil if you are resting a big roast, and understand the trade — a tight wrap keeps it hotter and softens the crust you spent effort building. For a steak, no cover at all is usually the better call.
Where to put the probe
A reading is only as good as its location. Aim for the thickest part of the muscle, and stop short of bone, gristle and fat pockets, all of which conduct differently from muscle and will read wrong in both directions. On a thin cut, come in from the side along the length rather than down from the top, which is the only way to keep the sensor inside the meat instead of poking through into the pan.
On a whole bird, the two places that matter are the thickest part of the breast and the inner thigh near but not touching the joint. The thigh runs behind the breast, and cooking until the thigh is done tends to leave the breast dry, which is why spatchcocking and separating the parts are both popular answers. Take the reading in more than one spot on anything large, since ovens have hot sides and a turkey can be 10 °F apart between the two thighs. The pop-up plastic indicator in a supermarket bird triggers well past done; ignore it.
Calibrate the thermometer occasionally: in a glass packed with crushed ice and topped with water, stirred and left a minute, it should read 32 °F. Many models have an adjustment for the offset, and the ones that do not at least let you know what to mentally subtract.
Pork, fish and the numbers that changed
Whole-muscle pork was recommended at 160 °F for decades because of trichinella, and that recommendation was revised downward to 145 °F with a 3-minute rest for chops, roasts and loin — the modern US pork supply made the higher figure unnecessary. A great many home cooks are still working from the old number, which is most of why pork loin has a reputation for being dry. Ground pork stays at 160 °F for the grinding reason above.
Fish is listed at 145 °F, though many cooks pull salmon and tuna well below that for texture, which is the same preference-with-risk category as a rare steak and belongs to the same caveats. Shellfish give visual cues instead: clams, mussels and oysters open, shrimp turn opaque and firm, and scallops go from translucent to milky. Discard shellfish that stay shut after cooking.
The one-sentence version
Buy a thermometer, put it in the thickest part, and treat any number below the published minimum as a choice you are making rather than a fact about the food. If you want a refresher on everything that happens to the meat before it reaches the pan, food safety at home is the companion piece, and fridge storage times covers how long you had before you cooked it.
Questions people ask
Is medium-rare steak safe?
It is below the USDA-recommended safe minimum of 145 °F, so the accurate answer is that it is a preference carrying a real risk rather than a safe choice. On an intact, whole-muscle cut, contamination is a surface phenomenon and a hard sear takes the surface far past any pathogen threshold, which is why the risk is low for a healthy adult. It is not low for someone pregnant, immunocompromised, very young or elderly, and those groups should cook to 145 °F. The reasoning also collapses entirely on ground meat and on cuts that were mechanically tenderized or injected, because those processes carry surface organisms into the middle.
Why does ground beef need 160 when steak only needs 145?
Because grinding redistributes the surface throughout the meat. An intact muscle is essentially sterile inside, so searing the outside handles it. Once meat goes through a grinder, what used to be the outside of a large batch is now mixed evenly through every burger, and there is no interior that was never exposed. The USDA figure for ground beef, pork, lamb and veal is 160 °F, and the same caution applies to blade-tenderized or brine-injected cuts, which are not always labelled in a way you would notice.
Do I have to cook chicken to 165?
That is the USDA safe minimum for all poultry in every form — whole birds, parts, ground, and stuffing cooked inside the cavity. Pathogen reduction is a combination of temperature and holding time, and 165 °F is the instantaneous figure chosen because it works without anyone needing to track seconds. Take the reading in the thickest part of the breast and in the inner thigh away from the bone, ignore the pop-up indicator that came in the bird, and ignore the color of the juices, which is unreliable in both directions.
How much does meat keep cooking after I take it off the heat?
It scales with mass. A large roast typically climbs another 5 to 10 °F and sometimes more, a thick steak 3 to 5, and a thin cutlet or a boneless chicken breast only a degree or two. That means a big roast has to come out below its target or it will sail past it. Resting is worth doing regardless of the temperature question, because heat drives moisture toward the center and a few minutes off the heat lets it redistribute instead of running out onto the board.
Can I tell doneness by pressing on it or by the juice color?
Not reliably. The touch test works for people who cook the same cut hundreds of times and calibrate against a thermometer, and it fails on unfamiliar thicknesses, on cuts with different fat content, and on anything cooked from a different starting temperature. Juice color is worse. Ground beef can brown before it reaches temperature and can also stay pink after it has passed it, depending on pH, oxygen exposure and whether it was frozen. Every one of these proxies is trying to estimate a number you can measure directly in three seconds.