The short answer: Weigh food raw when you can, and log it against a raw database entry. Cooking does not destroy calories. It moves water in or out, so the same food changes weight while its energy stays put. Raw weight is the steadier number because cooked weight depends on heat, time, and method. The real exception is fat: fat that drips out of meat, or oil that soaks into it, does change the calories you actually eat.

Does cooking change the calories or only the weight?

Mostly the weight. The USDA defines cooking yield as the change in food weight from moisture loss (evaporation or drip), water absorption during boiling, and fat gain or loss. The formula in the USDA Table of Cooking Yields for Meat and Poultry, Release 2 is simple. Yield (%) = 100 x (cooked weight / raw weight).

Notice what is not in that list. Heat does not burn off protein or carbohydrate. A cup of rice that swells to three times its dry weight still holds the same starch it started with. The USDA survey database documentation states the point directly. Water gained or lost during cooking "can have a substantial effect on the nutrient content when expressed on a per 100-gram basis" (FNDDS 2021-2023 documentation).

So the calorie number per 100 grams moves a lot. The calories in your portion barely move at all.

Does the USDA database list foods raw or cooked?

Both, as separate records. USDA FoodData Central carries "Rice, white, long-grain, regular, raw, enriched" and "Rice, white, long-grain, regular, enriched, cooked" as two different entries. Values in the Standard Reference data are reported as the amount in 100 grams, edible portion (SR Legacy documentation).

That gives you the rule that actually matters. Weigh the food in the state that matches the entry you chose. Raw weight with a raw entry is correct. Cooked weight with a cooked entry is also correct. The mistake is mixing the two.

One more detail explains why raw is often the better anchor. Many cooked entries were never analyzed in a lab. The USDA calculates them from the raw food by applying cooking yields and nutrient retention factors, which cover 16 vitamins, 8 minerals, and alcohol. The raw record is usually the measured one.

Where does raw versus cooked logging go most wrong?

Dry foods that drink water. Per 100 grams, FoodData Central lists:

  • White long-grain rice: 365 calories raw (11.6 g water) against 130 calories cooked (68.4 g water).
  • Enriched pasta: 371 calories dry (9.9 g water) against 158 calories cooked without added salt (62.1 g water).
  • Black beans: 341 calories raw (11.0 g water) against 132 calories boiled without salt (65.7 g water).

Log 100 grams of cooked rice against the raw entry and you record 365 calories instead of 130. That is a 235 calorie error from one side dish. Do it twice a day and the whole plan stops working. Our guide to how accurate calorie counts really are covers the other sources of drift.

Meat runs the opposite way, because it loses water instead of gaining it. Skinless boneless chicken breast is 120 calories per 100 grams raw and 165 calories per 100 grams roasted. Weigh it raw and log the roasted entry and you overcount by about a third.

When does cooking really change the calorie count?

When fat leaves the food or enters it. The USDA cooking yields table reports fat change as a percentage of the raw weight, and the numbers are not small:

  • Ground beef patty, 12 to 22 percent fat, broiled or grilled: 69 percent yield, water down 25.1 percent, fat down 5.17 percent of raw weight.
  • Ground beef patty above 22 percent fat, broiled or grilled: 63 percent yield, fat down 11.98 percent of raw weight.
  • Bacon, pan-fried: 32 percent yield, water down 36.8 percent, fat down 28.6 percent.

FDA rules let calories be calculated with general factors of 4, 4, and 9 calories per gram for protein, carbohydrate, and fat (21 CFR 101.9(c)(1)(i)(B)). At 9 calories per gram, roughly 12 grams of fat leaving 100 grams of raw high-fat ground beef is about 110 calories that never reach your plate. Raw weight alone counts them anyway.

The same logic runs in reverse when you fry. Oil absorbed by the food is real added energy, so weigh the oil that goes in the pan and log it separately. Lean cuts barely move: pork tenderloin roasts at an 80 percent yield with a fat change near zero, so the water is the only thing that left.

Are food labels based on raw or cooked weight?

As sold. FDA rule 21 CFR 101.9(b)(9) states that "the declaration of nutrient and food component content shall be on the basis of food as packaged or purchased." The exceptions are narrow. They cover raw fish and foods packed in a liquid that is not eaten. The 200 calories on a pasta box refer to dry pasta.

Manufacturers may add a second column under 101.9(e) that shows the same food "as purchased" and "as prepared." Cake mix and instant oatmeal often do this. Read the column header before you copy the number.

Meat and poultry follow USDA FSIS rules instead. 9 CFR 317.309(b)(3) requires the declaration to be "as packaged." One exception applies. Single-ingredient raw products that are not ground or chopped may be declared "as consumed." So a steak label may already be on a cooked basis while the ground beef next to it is raw. Check the panel rather than assuming.

How large is the error in real terms?

Match the state and the error nearly vanishes. Take raw chicken breast at 120 calories per 100 grams. Roasting a broiler-fryer breast leaves about 72 percent of the starting weight. Those 72 grams at 165 calories per 100 grams work out to about 119 calories. Same food, same energy, different scale reading. That also explains why cooked protein looks denser per gram, a point worth keeping in mind when you set targets using our protein intake guide.

Mismatch the state and the error is large and one-directional. Rice logged wrong runs 180 percent high. High-fat ground beef logged raw runs roughly 100 calories per 100 grams too high after the fat drips away.

Keep it in proportion, though. In a classic New England Journal of Medicine study, adults who could not lose weight underreported their food intake by 47 percent on average (Lichtman and colleagues, 1992). Forgotten foods and eyeballed portions still beat raw-versus-cooked as a source of error. Picking a method and holding to it matters more than picking the perfect one, which is the theme of our comparison of calorie tracking methods.

The bottom line

Weigh raw, log raw, and you have the most stable number available, because raw weight does not depend on how long you left the pan on. If you can only weigh cooked food, that is fine, as long as you select the cooked entry in your app. Track fat separately when it drains off or soaks in, since that is the only step where calories genuinely leave or join the food. Small consistent habits like this are what make a nutrition plan hold up over months, and you can read more about building them at thrive.md.

Last updated: September 2026. This article is for informational purposes only and does not constitute medical advice. Talk with a registered dietitian or your clinician before making major changes to your diet, especially if you manage diabetes, kidney disease, or an eating disorder.