How to Count Calories in Homemade Food Without a Label
What is the short answer?
Estimate the parts, not the plate. Break the dish into the three or four things that actually carry the calories — the starch, the protein, the fat it was cooked in, the sauce — put a number on each one using a portion you can recognise by eye, and add them up. Guessing a single total for "chicken and rice" is a coin toss; adding up four components you half-know is reliably close.
The second half of the answer matters just as much: estimate the same dish the same way every time. A record that is consistently a bit high is useful. A record that is high on Monday and low on Thursday tells you nothing.
Which components actually matter?
Most home-cooked meals are dominated by a small number of ingredients. Vegetables, herbs, spices and most seasonings are rounding errors. The things worth counting are:
The starch: rice, pasta, bread, potatoes, couscous, noodles, flatbread.
The protein: meat, fish, eggs, pulses, tofu, cheese.
The fat: oil, butter, cream, coconut milk, the mayonnaise in the dressing.
The obvious extras: nuts, dried fruit, sugar in a dessert, alcohol in a drink.
If you catch those four you have caught the large majority of the calories in the meal. Adding a careful figure for the onion is precision you do not need and cannot achieve.
What reference portions should I memorise?
You need a handful of anchors you can hold against a plate without a scale. These are ordinary published figures, rounded:
A level tablespoon of oil or butter is about 120 kcal. This is the single most useful number in cooking.
A fist-sized serving of cooked rice (roughly a cupful, 180–200 g) is about 250 kcal. The same volume of cooked pasta is a bit less, about 200 kcal.
A palm-sized piece of cooked lean meat or fish — roughly the size of a deck of cards, 100–120 g — is about 150–250 kcal depending on how fatty it is and how it was cooked.
A medium slice of bread is about 80 kcal; a thick-cut slice nearer 100.
A medium potato is about 130 kcal boiled, and considerably more once it has met oil.
A matchbox of hard cheese (about 30 g) is around 120 kcal.
A cupped handful of nuts (about 30 g) is around 180 kcal.
A cupful of cooked beans or lentils is around 230 kcal.
A tablespoon of mayonnaise is around 90 kcal; an oil-based dressing, around 70–80.
Five or six of these, learned properly, will cover most of what you eat. Check them once against a scale at home so your idea of "a fist" is calibrated to your own fist, and then stop weighing.
How do I account for the oil I cannot see?
Separately, and deliberately, because it is invisible in the finished dish and it is usually the largest single thing people miss.
A pan of vegetables fried at home might take one to two tablespoons of oil — 120 to 240 kcal that simply do not appear on the plate as anything you can point at. Roasted potatoes are glossy for a reason. A stir-fry sauce, a curry finished with cream, a salad that arrived already dressed: all of these carry fat that looks like nothing.
So make it a habit: after you have counted the starch and the protein, ask "what fat was this cooked in, and how much of it ended up in my share?" If you poured two tablespoons into a pan feeding four people, you ate roughly half a tablespoon — about 60 kcal. If a good deal of oil stayed in the pan, count less. If the dish is one that absorbs fat rather than sitting in it, count most of it.
When you are logging by sentence rather than by database entry, say the fat out loud: "chicken thigh, cup of rice, fried in about a tablespoon of oil" gives a much better estimate than "chicken and rice". Day So Far reads the sentence as written, so the oil you mention is the oil that gets counted.
How do I log a shared pot or a composite dish?
Estimate the whole thing, then take your fraction. This is far easier than trying to estimate your portion directly, because you know what went into the pot.
Say you made a stew with about 800 g of meat, a kilo of potatoes, three tablespoons of oil and a tin of tomatoes, and it fed six people. You can put a rough number on each of those, total it, and divide by six. If you had a generous plate, call it a fifth instead of a sixth. That single adjustment — the fraction — is where nearly all of your judgement goes, and it is a judgement you can actually make: you can see whether your bowl was bigger than everyone else's.
The same trick works for a cake, a tray bake, a large pizza, a jug of a mixed drink. Estimate the object, divide by the slices.
What about food someone else cooked?
At a friend's house you have no ingredient list, so fall back on comparison. Look at the plate and ask: how does this compare to the version I make? Half again as much rice? Twice the meat? Scale your own known estimate by that ratio and take the fat upward, because other people almost always cook with more of it than you assume.
Restaurants deserve a standing adjustment. The same dish in a restaurant tends to be both larger and richer than the home version — more butter, more oil, a bigger plate, bread and a sauce you did not ask for. When you are estimating a restaurant meal, do not take your usual figure. Take the top of your range, not the middle.
How do I handle the fact that I am guessing?
By guessing twice. Work out a plausible low figure and a plausible high one — "somewhere between 600 and 850" — and log the midpoint, 725. This does two things. It stops you anchoring on the number you would prefer, and it tells you when your estimate is too vague to be worth keeping: if the gap between low and high is more than about a quarter of the total, break the dish into smaller components until it narrows.
Then accept the error and move on. Estimates of unpackaged food are not measurements and will not become measurements no matter how long you stare at the plate. What you are building is a consistent signal, not an audit.
Why does consistency matter more than accuracy?
Because a log is only useful in comparison with itself. If your estimate of the same dish is steadily 10% low, every week is 10% low, and the trend — up, down, flat — is still correct. The moment you estimate generously on good days and strictly on bad ones, the trend dissolves and you cannot tell a real change from a change in your own guessing.
This is also why a tool that compares your logged intake against what the scale actually does over a fortnight can work with imperfect numbers: a steady bias gets absorbed into the calibration, while a wandering one never does. So write your estimates down the way you made them, keep a note of the method for dishes you eat often, and reuse it. Boring and repeatable beats clever and occasional, every time.