Data & method

The composition data

Nutrient values for single ingredients — wheat flour, rice, pulses, paneer, milk, oils, vegetables, meat, eggs — are taken from the इंडियन फ़ूड कंपोज़िशन टेबल्स 2017, published by the National Institute of Nutrition (ICMR) in Hyderabad. IFCT 2017 is the current authoritative analysis of Indian foods, measured on Indian samples, and it is the right base for Indian cooking in a way that a US database is not: Indian wheat, Indian pulses and Indian milk are not compositionally identical to their American equivalents.

Citation: Longvah T, Ananthan R, Bhaskarachary K, Venkaiah K. इंडियन फ़ूड कंपोज़िशन टेबल्स 2017. National Institute of Nutrition, Indian Council of Medical Research, Hyderabad. Nutrient values are measured scientific quantities, reproduced here as data with attribution; the publication's own text is not reproduced.

Composite dishes

Most of what people actually search for is not an ingredient — it is a dish. Biryani, chole, dal makhani and pav bhaji do not appear in any composition table because they are recipes, and every household cooks them differently. Those entries are built by standard recipe analysis: a normal home recipe is written out ingredient by ingredient, each ingredient's composition is taken from IFCT, water loss or gain during cooking is applied, and the result is divided by the finished weight.

Where a dish is normally restaurant food rather than home food — butter chicken, a biryani plate, pav bhaji — the recipe used is a restaurant-style one, with the fat that implies, and the page says so. Where a dish is normally home food, a measured home hand with the oil is assumed. Neither assumption is hidden: the serving box at the top of every page states it.

Serving weights

This is the part other sites skip. A calorie figure per 100 g is useless if you do not know what a roti weighs, so every dish here carries a stated serving in grams:

• One medium phulka roti — 40 g, made from about 25 g of atta, no ghee.
• One katori of cooked dal — 150 g, the standard katori used in ICMR–NIN dietary guidance.
• One katori of cooked rice — 150 g, roughly 50 g of raw rice.
• One restaurant plate of biryani — 350 g.
• One cup of masala chai with milk and sugar — 150 ml.

These weights come from ICMR–NIN dietary guidance for Indians and from weighing normal portions. They are typical, not universal. If your roti is bigger, scale the number — the arithmetic is linear and the per-100 g table on every page lets you do it in one step.

The energy check

Energy is not an independent measurement here; it has to agree with the macronutrients that produce it. Every build runs each food through Atwater factors — 4 kcal per gram of protein, 4 per gram of available carbohydrate, 9 per gram of fat — and refuses to publish if the stated energy is more than 15% away from that sum (20% for high-fibre foods, where a share of the carbohydrate is not fully available). A dataset of 246 foods entered by hand will contain typos; this check is what stops one becoming 246 wrong pages.

Carbohydrate is quoted as उपलब्ध carbohydrate, with dietary fibre listed separately, following the IFCT convention. That is why the columns on a food page do not always add to the serving weight.

Burning it off

The activity minutes on each page use the standard ACSM relationship, kcal per minute = MET × 3.5 × body mass in kg ÷ 200, for a 70 kg adult, with MET values from the 2024 Adult Compendium of Physical Activities. The full activity dataset is published on a sister site, BurnLab.

What these numbers cannot tell you

• Oil is the biggest variable, by far. One extra tablespoon of oil in a pan of sabzi is around 120 kcal spread across the dish. Two cooks making "the same" aloo sabzi can differ by 40%.
• Restaurant portions are larger. Where a plate size is stated it is a normal restaurant plate, but portions vary by city, by restaurant and by whether it arrived in a thali.
• Recipes are regional. Sambar in Chennai is not sambar in Bengaluru. The recipe used is a common one, not the only one.
• These are averages. Composition tables report a mean across analysed samples. Individual batches vary.
• Ghee is never assumed unless stated. If you apply ghee to your roti, add about 45 kcal per teaspoon yourself.

सिर्फ़ सामान्य जानकारी — चिकित्सकीय या पोषण संबंधी सलाह नहीं। घर के बने भारतीय खाने में तेल, घी और हिस्से के आकार से बहुत फ़र्क़ पड़ता है; यहाँ हर मान एक बताई गई सर्विंग का औसत है, और उसके साथ मानी गई बात भी दिखाई गई है।

Where each number comes from — the July 2026 audit

Carbohydrate can be reported two ways. By difference is whatever is left after water, protein, fat and ash are subtracted, and it still contains the dietary fibre. Available carbohydrate has the fibre taken out. Publishing a by-difference figure in a column labelled available overstates the carbohydrate, and once energy is reconciled against it, the calories as well — and because the two wrong numbers then agree with each other, the Atwater check above cannot see the error. Only an external table can. IFCT 2017 publishes available carbohydrate directly, which is why it is the base used here.

In July 2026 every single-ingredient food on this site was read back against IFCT 2017 to find rows where that had happened. Three were found and corrected: besan, suji और pistachios. Four potato dishes were caught by a different test — a katori cannot contain more potato than it weighs — and were recomputed against IFCT F006: jeera aloo, dum aloo, aloo chaat और aloo tikki.

Every food now carries a label saying how far this site can stand behind it, printed on its own page:

• सत्यापित — read off a named IFCT 2017 food code, and the published figure matches it. 13 foods.
• रेसिपी विश्लेषण — built from a stated ingredient list, each ingredient taken from a named table row, divided by the finished weight. The ingredient list is printed on the page. 54 foods.
• From the composition tables, not re-checked — the hand-entered composite dishes, plus a handful of single ingredients where IFCT was found but disagrees with what is published. Where that happens the IFCT figure is stated on the page rather than the row being quietly rewritten, because that is a change of source policy and not a correction. 175 foods.
• Not in any composition table — sugar, honey, toned milk and curd have no IFCT entry at all. 4 foods.

That third group is the honest limit of this site. 179 of 246 foods are still hand-entered composites with no stored ingredient list, which means the carbohydrate basis behind them cannot be checked from the outside — and those pages say so rather than implying a citation they do not have. The widest single unresolved gap is almonds, where IFCT gives 3.04 g of available carbohydrate and this site publishes 11 g.

The recipe engine, and what it is not

Also in July 2026, 54 of those composites were converted from typed-in numbers to stored ingredient lists. Their energy and macronutrients are now computed at build time from the ingredients, and the ingredient list and finished weight are printed on the page so the arithmetic can be redone rather than trusted. The point is not that the computed number is more accurate. The point is that it is checkable, and that correcting an ingredient now necessarily corrects every dish made from it — which a typed-in number can never do.

Two limits are worth stating plainly. First, energy on a computed row is the Atwater sum of its own ingredients, so it reconciles with the macro check above by construction. That is a property of the method, not evidence the dish is right. Second, the ingredient quantities are this site's assumption; only the composition of each ingredient is cited. Every converted dish therefore carries a confidence, and anything deep-fried or reduced over heat carries a low one, because the oil uptake or the reduction — not the ingredients — is what decides its energy.

Conversion was deliberately partial. A dish was converted only where its ingredients are dominated by a row with a named table code. Everything built on a dal, a chickpea or a sprout was left alone: IFCT's pulse codes are not established in this dataset, and guessing one to make a dal look computed would be worse than leaving it honestly labelled. Onion is the single most useful ingredient row this site still lacks — it is in most gravies and in the everyday salad, and its absence is why several otherwise straightforward dishes are still hand-entered.

सुधार

If a figure here disagrees with IFCT 2017, IFCT wins and the page gets fixed. Send the page URL and what looks wrong.

स्रोत·IFCT 2017 + प्रकाशित रेसिपी वज़न·समीक्षित जुलाई 2026

स्रोत·भारतीयों के लिए ICMR–NIN का आहार मार्गदर्शन + तौले गए हिस्से·serving weights — roti 40–45 g, katori 150 g, biryani plate 350 g·समीक्षित जुलाई 2026

स्रोत·2024 Adult Compendium of Physical Activities·MET values behind every burn-off figure·समीक्षित जुलाई 2026