Glycemic Index and Load Explained

How glycemic index is actually measured

GI is not calculated from a food’s composition. It is measured in people, which is why the numbers are less precise than the tables make them look.

The protocol: volunteers eat a portion of the test food containing exactly 50 g of available carbohydrate — total carbohydrate minus fibre. Their blood glucose is measured repeatedly over the next two hours and the area under the resulting curve is compared against the same person eating 50 g of pure glucose, which is defined as 100. The result is averaged across subjects.

Two consequences follow from that design, and both are routinely overlooked. First, GI describes a 50 g carbohydrate portion, not a realistic serving — you would need about 700 g of watermelon to hit 50 g of carbohydrate. Second, because it is measured in human subjects, published values carry real confidence intervals. The same food varies by variety, ripeness, growing conditions, processing and cooking method.

The category bands

Both scales are anchored to glucose at GI 100. Our GI calculator and GL calculator work with these bands.

Why glycemic load exists

Glycemic load was introduced to fix the portion problem. It scales the index by how much carbohydrate is actually in the serving you eat.

Watermelon is the standard illustration. Its GI is around 72 — high, on paper. But a 120 g serving contains roughly 6 g of available carbohydrate, so its glycemic load is (72 × 6) ÷ 100 = 4.3, comfortably in the low band. The index says the carbohydrate in watermelon acts fast; the load says there is very little of it in a normal portion. Both are true, and only the second tells you what will happen when you eat some.

GI values from international tables; carbohydrate figures are typical and vary by variety and preparation. Rows are rounded to one decimal.

What this does and does not tell you about insulin

The point that matters most if you dose insulin: neither GI nor GL tells you how much insulin to take. That comes from the carbohydrate count and your insulin-to-carb ratio, and nothing on this page changes it.

What GI can inform is timing. A high-GI meal raises glucose faster, so it may benefit from a longer pre-bolus; a low-GI, high-fat meal releases carbohydrate slowly and may spike hours later, when a normal bolus has already finished working. That is a timing and split-dose question rather than a dose-size one — our pre-bolus timing guide covers the first case and the Warsaw method the second.

There is also a measure people reach for GI to answer that it was never designed for: how much insulin a food actually provokes. That is the food insulin index, a separate measurement, and it diverges from GI for protein-rich foods that raise insulin without raising glucose much. Our food insulin index calculator covers it.

Where the numbers get shaky

Mixed meals. GI is measured on a single food eaten alone. Add fat, protein and fibre — which is to say, add a normal meal — and gastric emptying slows, changing the curve considerably. The GI of the rice does not describe the rice-with-chicken-and-oil you actually ate.

Preparation. Cooking time, processing and even cooling change the figure. Pasta cooked al dente has a meaningfully lower GI than the same pasta boiled soft, and potatoes cooled after cooking form resistant starch that lowers theirs.

Individual variation. Different people produce genuinely different glucose curves from identical meals. Published GI values are population averages, so they describe a typical response rather than yours.

It is not a health scale. Table sugar has a lower GI than a baked potato. Ice cream sits below wholemeal bread. GI ranks how quickly the carbohydrate in a food raises glucose, and answers no question about nutritional value.

Limitations & Safety Notes

• GI is measured on portions containing 50 g of available carbohydrate, not realistic servings. Glycemic load exists to correct for that and is the more practical figure.
• Published GI values carry wide confidence intervals and vary by variety, ripeness, processing and cooking method.
• GI is measured on single foods eaten alone. Fat, protein and fibre in a mixed meal change the response substantially.
• Individual glucose responses to identical meals differ between people, so published values are population averages rather than predictions for you.
• Neither GI nor GL determines an insulin dose. That comes from carbohydrate counting and your insulin-to-carb ratio.

Frequently Asked Questions