Weight loss

Nutrition & Diet Plans

The Glycaemic Index of Indian Food, Measured

The glycaemic index of Indian food depends on the preparation more than the grain: measured values run from 27.3 for a pulse-and-fibre roti to 93.4 for a processed finger-millet flour. "Eat low GI" is unusable advice until you know where your rice, roti and dal fall on the scale.

In testing by ICMR-NIN, cooked rice came out at a glycaemic index of 78.23 and whole-wheat chapati at 65.66, both in ten healthy Indian volunteers against a glucose reference. Testing at the Madras Diabetes Research Foundation put whole-wheat roti at 45.1. The last two are the same cereal, measured 20 points apart. Below are the Indian measured values, the conditions behind each one, and the checks that tell you which numbers to trust.

Key Takeaways

  • ICMR-NIN measured cooked rice at GI 78.23 ± 4.24 and whole-wheat chapati at 65.66 ± 4.22, in ten healthy Indian volunteers.
  • MDRF measured whole-wheat roti at 45.1 ± 3.5 and a pulse-and-fibre "atta mix" roti 17.8 points lower, at 27.3 ± 2.2.
  • Dals clustered tightly, from 37.95 for Bengal gram to 43.64 for mixed dal.
  • The millet preparations below tested between 37 and 53, while a decorticated, heat-treated finger-millet flour in the same review tested at 93.4, so a millet label alone does not predict a value.
  • The ICMR-NIN glycaemic loads other than the one for rice assume a 50 g available-carbohydrate test portion rather than a household serving, and the published millet loads do not state the serving they assume.

Measured GI and glycaemic load values for Indian staples

Every value below was measured in Indian volunteers against a glucose reference, where glucose scores 100. On that scale the international bands put low at 55 or below, medium at 56 to 69 and high at 70 or above, while India's food regulator draws a stricter line for a low-GI claim on a label: a GI below 55. A GI value is a property of one tested recipe measured in one group of people, which is why who was tested, and how, matters as much as the number.

Rice

Source: ICMR-NIN, ten healthy Indian volunteers (2024 Indian dietary guidelines)

  • Cooked rice: GI 78.23 ± 4.24, glycaemic load 49.38 ± 2.67

Source: MDRF Chennai, 2012, healthy Indian volunteers (no glycaemic load published)

  • Sona Masuri: GI 72.0 ± 4.5
  • Ponni: GI 70.2 ± 3.6
  • Surti Kolam: GI 77.0 ± 4.0

Wheat

Source: ICMR-NIN, ten healthy Indian volunteers (2024 Indian dietary guidelines)

  • Whole-wheat chapati: GI 65.66 ± 4.22, glycaemic load 32.83 ± 2.11

Source: MDRF, 2010, 18 healthy Indian volunteers (no glycaemic load published)

  • Whole-wheat roti: GI 45.1 ± 3.5
  • Pulse-and-fibre atta mix roti: GI 27.3 ± 2.2

Dals

Source: ICMR-NIN, ten healthy Indian volunteers (2024 Indian dietary guidelines)

  • Red gram: GI 43.01 ± 4.93, glycaemic load 21.50
  • Green gram: GI 42.45 ± 4.05, glycaemic load 21.22
  • Bengal gram: GI 37.95 ± 5.73, glycaemic load 18.97
  • Masoor: GI 42.15 ± 3.26, glycaemic load 21.07
  • Mixed dal: GI 43.64 ± 6.98, glycaemic load 21.82

South Indian breakfasts

Source: ICMR-NIN South Indian series, 2022, ten volunteers, FAO/WHO method

  • Plain dosa: GI 79.39 ± 6.8, glycaemic load 39.69 ± 2.7
  • Onion dosa: GI 79.69 ± 5.9, glycaemic load 39.84 ± 4.8
  • Vada with sambar: GI 36.89, glycaemic load 18.44

Two caveats apply to these loads. The glycaemic load figures describe the 50 g available-carbohydrate test portion rather than a bowl served at home, so 32.83 for chapati is not an estimate of what a household serving delivers. Cooked rice is the one exception: a 50 g portion at GI 78.23 works out near 39, so the published 49.38 implies a portion closer to 63 g. And where a study published no load, as the testing of three Indian rice varieties did, no load was published for that food.

The 2022 ICMR-NIN series on South Indian breakfast foods shows how far values spread inside one cuisine: onion dosa tested at 79.69 while vada with sambar tested at 36.89, more than 40 points apart on the same breakfast menu. Plain and onion dosa were tested separately in ten volunteers and came out within a third of a point of each other, too close to rank one above the other.

How these values were tested

Most of the values above come from three Indian testing programmes that used closely similar conditions, which is what makes their values broadly comparable with each other:

  • ICMR-NIN, Hyderabad: Ten healthy Indian volunteers, glucose reference, 50 g available carbohydrate per test food, capillary glucose sampling and a 120-minute incremental area under the curve, reported as a mean plus or minus error, measured in people rather than estimated in a laboratory.
  • MDRF, Chennai, 2012: Healthy Indian volunteers, glucose reference, 50 g available carbohydrate, 120-minute incremental area under the curve.
  • MDRF, 2010: 18 healthy Indian volunteers, glucose reference tested three times, 50 g available carbohydrate, blood sampled at minus 5, 0, 15, 30, 45, 60, 90 and 120 minutes.

ISO 26642:2010, the international standard for this kind of testing, expects at least ten healthy subjects, duplicate tests of the reference food and a two-hour incremental area under the curve. Older six-person studies, and studies done in people already diagnosed with diabetes, therefore transfer less well to a healthy reader even when their values appear in the same list.

One conversion also applies before any comparison, by convention rather than by that standard: a value measured against white bread is multiplied by 0.71 to put it on the glucose scale. Variation between individuals eating the identical food remains substantial as well.

What changes the value for the same food

Preparation conditions changed the measured value in several Indian studies, by amounts those studies quantified. Replacing whole-wheat flour with a pulse-and-fibre atta mix lowered roti by 17.8 points. Rice measured across milling grades gave white rice 79.6, undermilled rice 73 and brown rice 57.6, with higher amylose content and parboiling both tracking lower values, although several of those figures were estimated from laboratory starch digestion rather than measured in people.

The cooling-and-reheating step is less settled than it is usually presented. Chilling cooked starch does raise resistant starch, but a 21-person trial of reheated rice found no significant GI difference despite resistant starch of 2.55 g per 100 g against 0.20 g per 100 g. In that trial the fridge changed the starch and not the measured glucose response.

Values for whole preparations exist where the whole preparation was tested, as the 2022 series did with vada and sambar, rather than being assembled from the values of their parts.

Measured values for millet preparations

Millet is a category rather than a value. Indian in-vivo reports collected in a review of millets in pre-diabetes give recipe-level figures that fall in the low band, alongside one flour value far above it:

  • Barnyard-millet upma, dehulled: GI 50
  • Barnyard-millet upma, dehulled and heat-treated: GI 41.7
  • Little-millet flakes: GI 52.11, glycaemic load 9.24
  • Bajra cheela: GI 44.94, glycaemic load 18.06
  • Ragi and little-millet dosa, dumpling and roti: GI 37 to 53, glycaemic load 11.05 to 18.43
  • Finger-millet flour (decorticated, hydrothermally treated): GI 93.4, glycaemic load 47

The spread inside this short list is wide enough to make "switch to millets" an incomplete instruction. Finger millet tested at 93.4 in eight volunteers where it had been decorticated, hydrothermally treated and built into a diabetic food formulation.

That high value traces to the decortication and the fibre it strips out. Finger millet cooked as dosa, dumpling or roti tested between 37 and 53. The processing is what the number describes, not the grain. Serving bases and error margins differ between the original studies, so a pooled millet average cannot substitute for the value of the recipe in front of you.

Can two published glycaemic-load figures be compared?

Usually not. A load figure means something only against a stated serving, and the published Indian figures do not use a common one. The ICMR-NIN loads above, rice aside, assume the 50 g available-carbohydrate test portion fed in the study: 32.83 for chapati, and 18.97 to 21.82 across the five dals.

The millet loads, from 9.24 for little-millet flakes to 18.43 for a ragi preparation, do not state the serving they assume at all. A load of 9.24 is not evidence that little-millet flakes are gentler per serving than a dal at 21.07, because the millet studies do not state the serving behind their figures while the dal figure describes a 50 g test portion.

A load figure is also only as good as the weight behind it. Whether a staple is weighed raw or cooked changes that weight before any GI is applied. For roti, weighing the raw atta gives the more accurate figure. Before using a published load, find the serving it assumes. Where the original study states none, the figure cannot be used for comparison.

Which Indian foods have no measured value?

Several foods people ask about repeatedly have no published Indian value measured in people:

  • Poha
  • Plain maida products
  • Ordinary jowar and bajra rotis as cooked at home
  • Rajma as normally served
  • Polished and parboiled versions of the same rice tested as a pair
  • Common Indian fruits

Two documents often stand in for these missing values. The Indian Food Composition Tables of 2017 are a nutrient-composition catalogue rather than a GI list, and the 2024 Indian dietary guideline annexure reports values drawn from individual studies rather than a universal value list. Neither is an official national GI catalogue.

This is why a published Indian value has to be traced back to the study that measured it rather than to a national table. So when a confident Indian fruit or rajma value appears, label it imported or unverified rather than Indian. Treating a borrowed figure as a local measurement is one of the more durable Indian diet myths in circulation.

How to apply these values to what you actually cook

Judge the preparation, not the ingredient. Inside one MDRF study, whole-wheat roti tested at 45.1 and the same study's atta-mix roti at 27.3, a gap of 17.8 points from a change of flour blend alone. Three checks decide whether a number helps you:

  • Was it measured in Indian volunteers or imported from a foreign table?
  • Was it measured in people or estimated in a laboratory?
  • What serving does its load assume?

For the foods with no Indian value, a borrowed number is a guess, and how your own glucose responds to a repeated meal is the one reading specific to you. That is a conversation for a registered medical practitioner or registered dietitian rather than something to settle from a table, particularly alongside what a normal blood-sugar range looks like for Indian adults.

Where a borderline blood-sugar result prompted the low-GI advice in the first place, a clinician-led metabolic assessment with Voy can put your own readings next to what you actually eat, instead of leaving you to match recipes against a published list.

Not sure how your usual meals are affecting your blood sugar? A Voy clinician can read your results properly. Book an assessment.

This article is for general information and education only and is not medical advice, diagnosis, or treatment. GLP-1 and other medications referenced are prescription-only and are appropriate only for certain people under the supervision of a qualified clinician. Do not start, stop, or change any medication based on this article. Please consult a registered medical practitioner about your individual circumstances. Information reflects what was available at the time of review and may change.

Related articles

Premium Star Icon

Join 25 Lakh members getting healthy

Doctor-backed weight loss support you can trust

4.7 / 5 (28,937+ reviews)

Trusted by 25 Lakh +
Customers

Still confused?

App Store
Download on theApp Store
Google Play
Get it onGoogle Play