How Badly People Estimate a Portion

A portion size is a number, and like any number it arrives with a method attached. When that method is a guess, the number is wrong by a predictable amount. The published work on portion estimation puts the average error at 20–30% low, with the error growing as calorie density rises. That means the foods where the mistake costs most—oils, nut butters, cheese—are the ones people estimate worst.

This page is about food data: where the numbers on a label or in a database come from, what the rules permit them to say, and how to recover a figure that is closer to the truth. It is not medical advice. If you are managing a clinical condition, or if tracking intake is affecting your relationship with food, stop and talk to a clinician. In the US, the National Eating Disorders Association helpline is 1-800-931-2237; in the UK, Beat is 0808 801 0677.

Where a nutrition figure comes from

A calorie value is not measured directly in most cases. It is calculated from the macronutrient content using Atwater factors: 4 kcal per gram of protein, 4 kcal per gram of carbohydrate, 9 kcal per gram of fat, and 7 kcal per gram of alcohol. These are averages. The true metabolizable energy of a specific food varies with its fiber content, its processing, and the individual. The Atwater system is a convention, not a constant of nature.

Labels carry a tolerance. In the US, the FDA allows a food to be off by 20% from the labeled value for most nutrients, and by 20% for calories. In the EU, the tolerance for calories is typically ±20% for foods other than supplements. A label is a legal statement, not a laboratory measurement.

Rounding lets a label print zero for something that is not zero. Under US rules, a food with less than 5 kcal per serving can be labeled as zero calories. A food with less than 0.5 g of fat per serving can be labeled as zero fat. If a spray oil says “0 calories” for a 1/4-second spray, the actual amount may be small, but it is not zero. The arithmetic to recover it is simple: if the product is pure oil, at 9 kcal per gram, then 0.4 g of oil is 3.6 kcal, which rounds to 0. Four sprays is 14.4 kcal, which does not.

The size of the estimation error

The 20–30% figure comes from controlled studies where participants are shown a portion and asked to estimate its weight or energy. The error is not random. It is biased low, and the bias is larger for foods with higher energy density. A 2013 review in the British Journal of Nutrition (doi:10.1017/S0007114512004412) found that estimation error for amorphous foods—pasta, rice, cereal—was greater than for foods with a defined shape. A 2006 study in the Journal of the American Dietetic Association (doi:10.1016/j.jada.2006.07.004) reported that participants underestimated the energy content of a meal by about 20% on average, with larger underestimates for larger portions.

The practical consequence: if a person estimates a 2-tablespoon serving of peanut butter as 2 tablespoons but actually spreads 3, the error is 50% on that item. At 9 kcal per gram and about 16 g per tablespoon, 2 tablespoons is 288 kcal; 3 tablespoons is 432 kcal. The 144 kcal difference is not visible in the bowl.

Hand-size heuristics and their error

Hand-size guides are common because they require no equipment. They are better than nothing, but they are not precise. A palm-sized portion of protein is often given as 3–4 oz (85–115 g). A thumb-sized portion of fat is about 1 tablespoon (15 g). A cupped hand of carbohydrate is about 1/2 cup cooked. These are population averages. A large adult male’s palm may hold 6 oz; a small adult female’s may hold 2 oz. The error is easily ±50%.

A 2016 study in Appetite (doi:10.1016/j.appet.2016.02.033) compared hand-size estimates to weighed portions and found that the hand method was accurate within 25% for some foods and off by more than 50% for others. The method is a starting point, not a measurement.

What a scale changes

A digital kitchen scale costs very little and removes the single largest source of tracking error. It does not require trust in a label or a database; it measures the mass of the food. The arithmetic then uses the label’s per-gram values, which are themselves subject to the tolerances above, but the portion is no longer a guess.

A worked calculation shows how to recover a real figure from a label that uses rounding.

Label: "0 kcal per 1/4-second spray"
Product: pure oil, 9 kcal/g
Assume 1/4-second spray delivers 0.04 g oil
Step 1: 0.04 g × 9 kcal/g = 0.36 kcal
Step 2: 0.36 kcal rounds to 0 kcal under US labeling rules (<5 kcal per serving)
Step 3: 10 sprays = 0.4 g oil
Step 4: 0.4 g × 9 kcal/g = 3.6 kcal
Step 5: 3.6 kcal still rounds to 0 kcal per serving if serving is 1 spray
Step 6: But 10 sprays is 10 servings, so label must show 0 kcal × 10 = 0 kcal
Step 7: Actual energy = 3.6 kcal, not 0

The same arithmetic applies to a label that prints “0 g fat” for a food with 0.4 g fat per serving. If a person eats three servings, the actual fat is 1.2 g, which is 10.8 kcal from fat alone.

What people get wrong, and why

The mistake is not stupidity. It is a natural consequence of how food is presented. A tablespoon is a unit of volume, but the amount of food that fits on a tablespoon depends on how it is packed, whether it is level or heaped, and the temperature of the food. Peanut butter at room temperature spreads differently from peanut butter from the refrigerator. A “cup” of cereal depends on the shape and size of the flakes; a cup of uncrushed cornflakes weighs about 28 g, while a cup of granola weighs about 110 g. The label’s serving size is a regulatory construct, not a recommendation.

People also anchor on the label. If the label says a serving is 2 tablespoons, the assumption is that the amount used is 2 tablespoons. The actual amount is rarely measured. The error is invisible because there is no feedback.

A table of conditions that change the figure

What to do differently

Weigh the calorie-dense foods. Oil, nut butter, cheese, butter, and salad dressing are the items where a small volume difference produces a large energy difference. A scale resolves the portion in grams, and the label’s per-gram value—even with its 20% tolerance—is a better anchor than a guess.

For foods that are not calorie-dense, a scale is still useful but the error matters less. A 20% error on a cup of broccoli is about 6 kcal; a 20% error on a tablespoon of oil is about 24 kcal. The same percentage error costs more where the energy is concentrated.

Read the label’s serving size as a legal unit, not a suggestion. If the label says 2 tablespoons and the actual amount is 3, the arithmetic is 3/2 times the labeled calories. That is the only correction that matters.

If tracking intake is causing distress, or if you are unsure whether your intake is appropriate for your situation, consult a registered dietitian or physician. This page is not medical advice. In the US, the National Eating Disorders Association helpline is 1-800-931-2237; in the UK, Beat is 0808 801 0677.