Understanding the Sydney University GI Database (And How Glyc Uses It)
Every glycemic load calculation starts with a glycemic index value. The most authoritative source of those values in the world is the International Glycemic Index Database maintained by researchers at the University of Sydney in Australia. Understanding where these numbers come from — and their limitations — helps you interpret them correctly.
What the database contains
The Sydney University GI Database catalogs glycemic index values for over 4,000 foods tested in controlled human studies. It's been maintained and updated since 1995, with major revisions published in 2002, 2008, and 2021. Each entry represents a food tested according to a standardized protocol in human subjects — not estimates, not calculations, not extrapolations.
The database covers common foods (white bread, rice, pasta, fruits, vegetables), branded products (specific cereal brands, snack bars, breads), and regional foods from around the world. Australian, North American, European, and Asian foods are all represented, though coverage varies by cuisine.
How GI testing works
Measuring a food's glycemic index requires a standardized protocol with human subjects. Here's how it works:
Select 10 or more healthy volunteers — the standard requires at least 10 subjects to account for individual variation. Some studies use more.
Establish a reference baseline — each subject consumes a reference food (either pure glucose or white bread) containing 50 grams of available carbohydrates. Blood glucose is measured at regular intervals over the next 2 hours.
Test the target food — on a separate day, the same subjects consume the test food, also containing 50 grams of available carbohydrates. Blood glucose is measured on the same schedule.
Calculate the area under the curve (AUC) — for both the reference food and the test food, the total glucose response over 2 hours is calculated as the area under the blood glucose curve.
Compute the GI — the GI equals (AUC of test food / AUC of reference food) × 100. If glucose is the reference, it gets a GI of 100. If white bread is used, values are adjusted to the glucose scale.
A food with a GI of 70, then, raises blood glucose to 70% of the level that pure glucose would — when you eat the same amount of available carbohydrate from each.
Why different sources show different GI values
Look up the GI of brown rice and you might find values ranging from 50 to 72 depending on the source. That's not an error. Several legitimate factors drive the variation:
Variety — there are thousands of rice varieties worldwide. Basmati rice has a GI around 50; short-grain sushi rice sits around 72. Both get called "brown rice," but they're not the same food.
Growing conditions — the same variety grown in different climates and soils can have different starch compositions.
Processing — parboiled rice has a different GI than non-parboiled rice of the same variety. Cooking time, water absorption, and cooling all affect starch structure.
Subject variation — different study populations (different countries, ages, metabolic health) produce somewhat different results even with the same food.
Reference food — some older studies used white bread as the reference (GI = 100 for white bread), while newer studies use glucose (GI = 100 for glucose). The white bread scale produces higher numbers. Values should always be converted to the glucose scale for consistency, but not all sources do this correctly.
Testing methodology — capillary blood versus venous blood sampling, different glucose measurement devices, and different time intervals between measurements can all introduce variation.
This is exactly why the Sydney University database is valuable: it aggregates results from multiple studies, notes the testing conditions, and provides a standardized reference point.
How Glyc uses the database
When you extract a recipe in Glyc, each ingredient goes through a lookup process to find its GI value. Glyc cross-references ingredients against data derived from the Sydney University database and the USDA FoodData Central database. The lookup follows a priority order:
Exact match — if the ingredient name matches a database entry exactly, that value is used. This is the most reliable match.
Alias match — many ingredients have multiple names. "Aubergine" should resolve to "eggplant." "Garbanzo beans" should resolve to "chickpeas." Glyc maintains an alias table for common synonyms.
Partial match — if no exact or alias match exists, Glyc uses a scoring algorithm to find the best partial match. This handles cases like "organic baby spinach" matching against "spinach" in the database. The scoring algorithm is designed to prevent false matches — "water" should not match "watermelon," and "cream" should not match "ice cream."
Hardcoded fallback — for approximately 65 common foods, Glyc maintains hardcoded GI and nutrition values that are used when the database is unavailable. This ensures the calculator works even during database outages.
When a partial or approximate match is used, Glyc notes this in the Show the Math section so you can verify that the match is reasonable.
What happens when a food is not in the database
Not every ingredient has a published GI value. Spices, herbs, oils, and most condiments have negligible carbohydrate content and contribute essentially zero to the glycemic load — Glyc assigns them a GI of 0, which is functionally correct.
For uncommon ingredients that do contain carbohydrates but lack a GI value, Glyc uses the closest available match and marks it as approximate. In some cases, no reasonable match exists, and the ingredient is noted as "not found" in the calculation breakdown. That's transparent by design — you can see exactly which ingredients were matched and which were not.
Database update frequency
The Sydney University GI Database receives major updates roughly every 6 to 12 years. The most recent comprehensive update was published in 2021. Between major updates, individual studies publish new GI values that eventually get incorporated.
That slow update cycle is actually appropriate. GI values for whole foods don't change rapidly — a banana in 2021 has essentially the same GI as a banana in 2008. Branded products change more often as manufacturers reformulate, but the database reflects the tested version at the time of publication.
Glyc syncs its data with published values and updates as new research becomes available. For most users, the existing values are accurate and sufficient. The rare case where a value might be outdated is a reformulated branded product — and those are best checked against the manufacturer's current nutrition label.
From GI to GL: why Glyc goes further
The Sydney University database provides glycemic index values. Glyc takes that data and calculates glycemic load, which accounts for actual portion sizes. This distinction matters because GI alone can be misleading.
Watermelon has a GI of 76, which looks alarming. But a typical serving (about 120 grams) contains only around 9 grams of available carbohydrate. The glycemic load of that serving is approximately 7 — solidly in the low range. You'd need to eat over a pound of watermelon to get the same glycemic impact as a cup of white rice.
Carrots are another example. GI of 39 (moderate), but a serving of raw carrots has a GL of about 2 because the carbohydrate content per serving is low. Carrots were unfairly maligned for years based on an older — and since corrected — GI value of 92.
By calculating GL for complete recipes with real serving sizes, Glyc provides a number that corresponds to actual blood sugar impact rather than a theoretical ranking of isolated carbohydrate sources. That's the practical information people need for meal planning, and it starts with the rigorous GI testing the Sydney University research team has maintained for over three decades.