Estimating by eye is worse than weighing, and no article can make that untrue. What it can do is tell you where the error actually lives, which is not spread evenly across your plate.
Vegetables and lean meat survive a rough estimate. Oil, nuts, cheese and dressings do not, because a small misjudged volume carries a lot of energy. Guess the first group freely; slow down for the second.
The honest position
We have written three articles on this site that all end the same way: put it on the scale. That advice is correct and it is also, for most people on most evenings, not going to happen.
So here is the position this page takes, and it will not shift: estimating is worse than weighing. Anyone telling you the hand method is as good as a scale is either selling something or has not looked. But you have a fixed budget of effort, and the interesting question is not accurate-versus-inaccurate. It is where to spend the budget — because the error is not spread evenly across your dinner, and most people spend their effort in exactly the wrong place.
The hand method, and what its authors actually claim
The version nearly everyone cites comes from the coaching company Precision Nutrition. It uses your own hand as a portion guide, which has the real advantage of scaling with the person: a larger person has larger hands and needs more food.
- Palm — protein
- Fist — vegetables
- Cupped hand — carbohydrates
- Thumb — fats
They do publish some numbers, and the numbers are the interesting part. A palm of protein is given as roughly 20–30 g of protein, or about 85–115 g of cooked meat. A thumb of fat is given as roughly 7–12 g.
The number their imitators repeat is that hand portions are 95 percent as accurate as weighing and measuring. It comes from Precision Nutrition's own internal work: unpublished, with no metric, no sample and no method attached. We are not going to call it false, and we cannot call it verified either. What is worth more is the sentence most of the people quoting it leave out — that given the known error rates in labels and nutrient databases, they expect this level of accuracy to be enough for all but the most advanced individuals. It comes from a company whose coaching is built on the method, which makes the admission stronger rather than weaker.
How accurate is it, really?
This is the question everyone asks and almost nobody answers with a source, so we are going to spend some time here.
First, the study everybody cites has been retracted
For over a decade, the most-quoted work on how badly people estimate meal calories was a 2006 paper in the Annals of Internal Medicine. The journal retracted it on 4 December 2018, reporting that Cornell University's Office of the Vice Provost for Research had stated the author “committed academic misconduct in his research and scholarship”.
You will still find its conclusions repeated all over the internet, usually as a confident percentage. We are not going to repeat them, and if you see that number on a page, it tells you something about how carefully the page was written.
What the peer-reviewed evaluation found
A 2025 evaluation looked at a hand-scale method among more than a thousand free-living adults, using four-day weighed food records across roughly twelve thousand meals. The authors' own conclusion is the one to hold on to:
Group level, not individual level. That is the sentence. Across a thousand people, the hand method lands in the right place. For your dinner tonight, it does not, and the researchers say so themselves rather than us saying it for them.
Some caveats that honest reporting requires: those models included participant characteristics and not only the number of hands, they predicted grams rather than calories, the hand-to-food mapping was not identical to the popular version, and the cohort was Japanese.
The other study everyone cites tested something else
There is a well-travelled claim that hands are “80 percent accurate”, traced to a 2016 study. That study is real, and it found something worth knowing — but not the thing it is cited for. It tested a finger-width technique with geometric formulas, on foods that closely resemble geometric shapes, not the palm-fist-cupped-hand-thumb guide.
What it found, on that easier task: 80% of estimates fell within ±25% of the true weight, against 29% for household measures — and only 13% landed within ±10%, against 8% for household measures.
Both readings survive, and both are useful. Hands comfortably beat measuring cups. And even the better technique, on the easiest possible food, put roughly one estimate in eight inside ±10%.
And trained people are not much better
A study of 54 postgraduate dietetics students at three institutes in Delhi — people training to do this professionally — found a mean absolute error of about 28% (SD 27%) using household utensils. Their worst categories were fruit juice, and salads on the food-exchange method rather than utensils — not the ones you would guess.
The point of citing this is not despair. It is calibration: if trained estimators sit near a quarter off, an untrained one aiming for five percent is chasing something that does not exist, and the effort is better spent elsewhere.
Why it breaks: two errors multiplied
Now the mechanism, because knowing why tells you where to be careful.
Your hand fixes roughly one volume. Two conversions then stand between that volume and a calorie figure, and each one has its own error:
- Volume to mass. How much the food weighs for a given volume. Compare a cup of cornflakes with a cup of granola: same cup, and the granola weighs several times more. Nothing about your hand knows this.
- Mass to calories. How much energy is in that mass. A gram of cucumber and a gram of olive oil are not in the same conversation.
Those errors multiply rather than add. A modest misjudgement in each becomes a large one by the end — and it is why a chart of hand equivalents cannot rescue the method. The chart would have to know which food, in what state, at what density, before its number meant anything.
The ten-second check on your own shelf. US labels print the serving in a household measure and in grams. Take two boxes out of your cupboard — a flaked cereal and a granola or muesli — and compare the grams behind “one cup”. The gap is larger than anyone expects, and you will never trust a volume again.
The four that actually cost you
Rank foods by how much energy hides in a small misjudged volume, and the list is short:
| Food | Why the guess is expensive | What to do instead |
|---|---|---|
| Oil and butter | The densest thing in the kitchen, and poured rather than portioned. A tablespoon of oil is about 119 calories by our own arithmetic — misjudge half a tablespoon and you have missed roughly 60 | Weigh the bottle, not the food |
| Nuts and nut butter | A handful is a volume with an enormous mass range, and the energy density is near the top of the food chain | Count them out, or pre-portion once |
| Cheese and cream | Grated, poured or melted — all three destroy the visual cue you would have used | Read the block's own panel and cut by eye against it |
| Dressings and sauces | Poured over a bowl, where nothing gives you a reference for how much went on | Dress in a measured way, or on the plate rather than in the bowl |
Everything else — the chicken, the rice, the vegetables — tolerates a rough estimate far better than people fear. If you want the daily number all of this is measured against, the calorie calculator works it out without an account. If you spend your one unit of care on the oil rather than on the broccoli, you have bought most of the available accuracy for almost none of the effort.
How wide is your own band?
Three numbers. Nothing is stored, and no account is needed.
Borrow a scale for one week, then give it back
This is the recommendation this page exists to make, and it is not “buy a scale and use it forever”.
Most people eat the same eight or ten things repeatedly. You do not need to weigh food in general — you need to weigh those, once, and learn what your own hand actually holds.
- List your repeats. The breakfast, the two or three lunches, the handful of dinners, the snack. Eight to ten items covers most of a normal week.
- Weigh each one, once, as you would actually serve it. Not a recommended portion — yours. Note whether you weighed it raw or cooked and match the database entry to that: weighing cooked rice against a dry-rice entry is a threefold error, larger than everything else on this page put together. The finished-weight method exists to keep the two straight.
- Write the grams down somewhere you will find them again.
- Put the scale away. You are now estimating against something you measured rather than against a chart written for somebody else's hands.
The honest caveat. Calibration fades. Portions creep, and the hand that learned in January is generous by June. This is a habit to repeat occasionally, not an exam you pass once.
Eating out is the case where you have neither a scale nor your own recipe — restaurant calories are handled separately, because there the number is often already printed for you.
For the meals you cook in one dish and serve to several people, there is a better move than estimating at all: weigh the tray once and the piece you take, which costs about the same effort and removes the guess entirely.
Decks of cards and tennis balls
The other standard advice is to compare food to household objects. A deck of cards for three ounces of meat, a ping-pong ball for two tablespoons of peanut butter, a baseball for — and here it falls apart.
Some published cards say a baseball is a cup. Others say half a cup. That is a twofold disagreement inside the very tool that is supposed to remove the guesswork, and it is a fair reason to treat the whole genre with suspicion.
There is also a quieter problem with the classic one. A deck of cards is a volume; three ounces is a mass. The comparison silently assumes one density for all meat, which is the same mistake the hand method makes, just with a prop.
Use them if they help you picture something. Do not treat them as measurements, and do not let them make you feel you have measured.
Everyone is already wrong
One last piece of perspective, because this page could otherwise read as discouraging.
Self-reported food intake comes out low. Checked against doubly labelled water — the reference method — a typical 24-hour recall understates by around a tenth, and across the studies gathered in a 2019 systematic review the shortfall runs from roughly a tenth to more than a third depending on the instrument. The ranges are wide and we are not going to pretend they are narrower than they are, but the direction is not in dispute.
Which reframes the whole question. You were never choosing between an accurate log and an inaccurate one. You were choosing where to put a fixed amount of care inside a system that has error in it either way. Put it on the fats, keep your entries consistent, and judge the trend over weeks rather than the total on any day.
If two of your entries disagree and you want to know which to keep, the reasons two apps show different numbers is worth reading next.
The short version
- Estimating is worse than weighing. Nothing here changes that.
- The error is not spread evenly: oil, nuts, cheese and dressings carry it.
- A hand is a volume, and volume-to-mass-to-calories is two errors multiplied.
- The peer-reviewed verdict on hand portions is group level yes, individual level no.
- Weigh your eight repeat foods once, write down the grams, then estimate against your own numbers.
- An error that lands in both directions costs far less than one that always leans the same way.
Please read. This is a method for counting, not dietary advice. If you are managing a medical condition, pregnant or breastfeeding, or working to a therapeutic diet, use the figures your clinician gives you. If counting is making your relationship with food worse rather than better, stopping is a reasonable answer — in the US the National Alliance for Eating Disorders helpline is 1-866-662-1235. See our health disclaimer.
You can earn the full version instead of paying for it
If you are estimating anyway, a photo of the plate is a faster estimate than a chart of hand portions, and photo logging is on the free plan. It has the same blind spot the rest of this page describes — it cannot see the oil that went in before the picture.
- Create an account and log a few meals. New accounts open with a two-day Plus trial — three days if somebody invited you — and then settle onto the free plan: no card, no end date, one meal scan a day.
- Publish an honest post, story, write-up or short video about what you found — or invite three people who go on to actually use it.
- Send us the link. A published piece unlocks the higher tier; three qualified invites unlock 90 days of the full version.
Your post has to say you got free access in exchange — that is an FTC requirement in the United States. Access is granted whether what you write is positive or critical. Full terms: promotional access.