How this instrument works
Glycemic index (GI) ranks how quickly a gram of a food's available carbohydrate raises blood glucose relative to a reference food, on a scale that runs roughly from 0 to 100 or beyond. It says nothing about quantity. Glycemic load (GL) fixes that by multiplying GI by the grams of carbohydrate in an actual serving, then dividing by 100. The result folds speed and amount into one figure: a food can rank high on GI yet contribute a small glycemic load if the portion carries little carbohydrate, and a moderate-GI food can produce a large glycemic load if the portion is generous.
The concept was introduced in a 1997 Diabetes Care study by Salmerón, Ascherio, Rimm and colleagues, tracking diet and diabetes risk in a large cohort of men and finding that glycemic load, not glycemic index alone, tracked more closely with risk. Working GI values for individual foods come from a separate reference: Foster-Powell, Holt, and Brand-Miller's 2002 international tables in the American Journal of Clinical Nutrition, still the standard source most nutrition tools draw on today. Neither paper enshrines fixed load categories, but a low/medium/high split of roughly under 10, 11 to 19, and 20 or more is in wide practical use, popularized through secondary sources such as Harvard's Nutrition Source rather than stated as a formal cutoff in either original study.
The arithmetic itself is simple multiplication, which is exactly the point — it forces the two variables that matter, response speed and serving size, into a single comparable number. A slice of watermelon carries a high GI near 72 but only about 8 grams of carbohydrate per typical serving, so its glycemic load lands low; a large bowl of white rice carries a moderate GI but enough carbohydrate to push its load high. Comparing foods by GI alone would rank these backwards for real-world impact.
- Enter Glycemic Index (0-100) — look this up from a published GI table for the specific food.
- Enter Carbohydrate per serving (g) — the actual carb content of the portion you plan to eat, from a label or database.
- Read Glycemic load: under 10 is generally read as low, 11-19 as medium, 20 or above as high.
Worked example — GI 55, 30 g carbohydrate
A serving with a glycemic index of 55 and 30 grams of carbohydrate: 55 × 30 = 1650, divided by 100, gives a glycemic load of 16.5 — squarely in the medium band (11-19). Neither the index nor the portion alone is extreme; it is a middling food eaten in a middling amount.
Compare two foods that push the same arithmetic to opposite ends. A very high GI of 95 paired with only 10 grams of carbohydrate — a small taste, not a full portion — gives 95 × 10 / 100 = 9.5, a low glycemic load despite the alarming index, because so little carbohydrate is actually present. A moderate GI of 40 paired with a generous 60-gram carbohydrate portion gives 40 × 60 / 100 = 24, a high load despite the unremarkable index, because the sheer quantity of carbohydrate dominates the result. Same formula, opposite lessons.
Questions
What is the real difference between glycemic index and glycemic load?
Glycemic index measures rate per gram — how fast a food's carbohydrate raises blood sugar compared with a reference, independent of how much you eat. Glycemic load measures the total expected impact of an actual serving by multiplying that rate by the grams of carbohydrate present. A food can have a frightening GI and a harmless glycemic load if servings are naturally small, or a modest GI and a large glycemic load if servings are naturally large.
Where do the low, medium, and high glycemic load categories come from?
They are not written into either founding paper. The rough bands — low under 10, medium 11 to 19, high 20 and above — come from secondary sources, most visibly Harvard's Nutrition Source, and have simply become the practical convention nutrition tools use. Treat them as a widely shared rule of thumb rather than a clinical threshold with the same evidentiary weight as the original glycemic-load research.
Why does a watermelon have a high glycemic index but a low glycemic load?
Because watermelon is mostly water. Its GI runs high, in the 70s, since the sugar it does contain enters the bloodstream quickly — but a typical serving carries only a small amount of actual carbohydrate, roughly 8 grams. Multiply a high index by a small carbohydrate figure and the product, the glycemic load, stays low. This is the exact scenario glycemic load was designed to catch that glycemic index alone would miss.
Does glycemic load account for fat, protein, or fiber in a meal?
No — the formula uses only glycemic index and carbohydrate grams for a single food or serving. Fat, protein, and fiber all slow digestion and blunt the blood-sugar response of a mixed meal in ways this simple multiplication does not capture. Glycemic load is a useful comparison tool between carbohydrate sources, not a complete model of how an entire plate will affect blood sugar.
Is a daily total glycemic load meaningful, not just a per-food number?
Yes — the original 1997 research examined glycemic load summed across a person's whole diet, not single foods in isolation, and found the daily total tracked with diabetes risk more closely than any single food's index. Adding up the glycemic load of everything eaten in a day gives a rough daily figure some clinicians use as a broader dietary pattern indicator, though individual meal-to-meal blood sugar response still varies with far more than carbohydrate alone.
References
- Salmerón J et al. 1997, Diabetes Care — dietary fiber, glycemic load, and NIDDM risk (PubMed)
- Foster-Powell K et al. 2002, Am J Clin Nutr — international GI/GL tables (PubMed)
Read this first: This instrument computes a screening figure from population formulas — it is not a diagnosis, and it cannot see the whole picture a clinician can. Use it to inform a conversation, not to replace one.