Guides

How to Track Calories Without Weighing Your Food

You do not need a kitchen scale to track calories usefully. Hand portions, visual references, and the accuracy you actually need to hit a goal, explained.

By Bento Bunny Team
Bento Bunny estimating portion sizes without reaching for a kitchen scale

Food scales are accurate and most people abandon them within a fortnight. That is not a discipline failure, it is a friction problem, and there is a version of calorie tracking that survives contact with real life. Here is how to do it without weighing anything, and what accuracy you genuinely give up.

The short version

  • Your hand is a portable, body-scaled measuring set. Palm, fist, cupped hand, thumb.
  • Fats and calorie-dense foods need the most care. Being 20g out on broccoli costs 7 calories. On olive oil it costs 180.
  • Photo-based estimation handles the plate-level guess for you.
  • Accuracy without a scale runs 10 to 20% out. That is fine for fat loss, tight for a physique deadline.
  • Judge by the trend, not the estimate. If the scale is not moving in a month, your estimates are high. Adjust and move on.

The Hand Method

The reason hands work as a reference is that they scale with body size, so a larger person with larger hands gets proportionally larger portions. Convenient, and roughly correct.

Reference Roughly Typical calories
Palm of meat or fish100 to 120g150 to 250
Cupped hand of rice or pastaAround 1 cup cooked200 to 220
Fist of vegetables1 to 1.5 cups30 to 60
Whole thumb of oil or butterAround 1 tbsp100 to 120
Cupped hand of nutsAround 30g170 to 200
Two fingers of cheeseAround 30g100 to 120

Calibrate once if you can. Weigh your own hand portions on a scale a handful of times, note where you run high or low, then put the scale away. Ten minutes of calibration buys you months of decent estimates.

Spend Your Attention Where the Calories Are

This is the principle that makes scale-free tracking work: estimation error scales with calorie density.

Misjudge a portion of spinach by 50g and you are wrong by about 12 calories. Nobody cares. Misjudge peanut butter by 50g and you are wrong by nearly 300. Same error in grams, twenty-five times the consequence.

So be deliberately careful with:

  • Oils, butter, and cooking fats. The worst offender, because they are invisible once cooked.
  • Nuts, nut butters, and seeds. Dense, and almost always under-poured.
  • Cheese, cream, and full-fat dressings.
  • Anything liquid with calories. Juice, smoothies, alcohol, oat milk in coffee.

And be relaxed about vegetables, salad, lean protein portions, and broth-based soup. Rough is fine there.

Let the Camera Do the Estimating

The hand method asks you to convert what you see into grams and then grams into calories. AI photo estimation collapses both steps: you take one picture and get a number back.

It is not magic and it is not scale-accurate. It works by recognising the foods and estimating volume from the image, which is a genuinely hard problem for dense or layered dishes and quite reliable for a normal plate of identifiable things. In practice a photo estimate lands in a similar accuracy band to a careful hand estimate, and takes about two seconds instead of thirty.

The realistic best approach is both: photograph the plate, then sanity-check the one or two dense items the camera cannot see, like the oil the vegetables were roasted in. You can try this without installing anything using the free AI calorie estimator.

How Accurate Is Scale-Free Tracking, Really?

Careful estimation without a scale typically lands within 10 to 20% of the true value. Whether that is good enough depends entirely on what you are doing.

Goal Scale needed? Why
General fat lossNoThe trend corrects for consistent bias
Hitting a protein targetOccasionallyWorth calibrating your palm once
Building the tracking habitDefinitely notFriction is the thing that kills it
Contest prep or a physique deadlineYesSmall errors compound near the limit

Here is the part that makes it work. If your estimates are consistently 15% low, your logged 2,000 calories is really 2,300. That sounds like a disaster, but it is not, because it is consistent. You set your target from the trend rather than from theory: eat what you are eating, watch four weeks of weight data, and adjust the target until the trend goes the direction you want. The bias cancels out. What does not cancel out is randomness, which is what you get from logging some days and not others.

Building the Habit That Survives

  1. Log immediately. Not at the end of the day. Recall shrinks portions and deletes snacks.
  2. Log the imperfect entry. A rough guess in the app beats a perfect one in your head.
  3. Save your repeats. Most people eat variations of the same ten meals. Log them once properly, reuse forever.
  4. Do not chase daily numbers. Weekly averages are the unit that means anything.

The bottom line

You do not need a scale to track calories usefully. Use your hand as a reference, be careful specifically with fats and dense foods, be relaxed about vegetables, and judge your progress from the four-week trend rather than any single day's total. Bento Bunny removes most of the estimating anyway: photograph the plate or type what you ate in plain words, and the AI returns calories and macros in seconds, no scale and no database search. It is free to download and runs on-device on iOS 26 and later. If you keep losing the habit rather than the accuracy, read what to do when you stop logging for a few days.

Frequently Asked Questions

Can you track calories accurately without a food scale?
Accurately enough for most goals, yes. Careful estimation using hand portions or AI photo estimation typically lands within 10 to 20% of the true value. That is sufficient for fat loss and habit building because the error is consistent, which means the weekly trend still tells you the truth. Contest prep or a strict physique deadline is where a scale earns its place.
How do I measure portions with my hands?
Use your palm for meat and fish at roughly 100 to 120g, a cupped hand for rice or pasta at about one cooked cup, a fist for vegetables at one to one and a half cups, your whole thumb for a tablespoon of oil or dressing, and a cupped hand for around 30g of nuts. Calibrate against a scale a few times first so you know whether you run high or low.
Which foods do I need to be most careful estimating?
Calorie-dense ones, because estimation error scales with density. Oils and cooking fats, nuts and nut butters, cheese, cream, full-fat dressings, and calorie-containing drinks all punish a bad guess heavily. Being 50g out on spinach costs you 12 calories. Being 50g out on peanut butter costs nearly 300.
Is AI photo calorie tracking accurate enough?
It lands in a similar accuracy band to careful manual estimation, roughly 10 to 20% for a normal plate of identifiable foods, and it takes seconds instead of minutes. It struggles most with layered or mixed dishes and with fats absorbed during cooking, so it is worth sanity-checking the dense items it cannot see.
What if my estimates are always too low?
That is fine as long as they are consistently too low. Set your target from your own data rather than from a formula: log for four weeks, watch the weight trend, and adjust the calorie target until the trend moves the way you want. A consistent bias cancels out. Inconsistent logging does not.

Start tracking with Bento Bunny

AI calorie tracking — just Type what you eat.