Potatoes, Rice, and Why Glycemic Index Misleads

A potato eaten with chicken or tofu does not behave like a potato tested alone after an overnight fast. That distinction changes the usual potato-versus-rice story. In a randomized crossover trial, potatoes with protein produced lower post-meal blood glucose than white rice with protein, even though the potatoes had higher isolated glycemic index (GI) values than the rice. The 2024 trial by Amr et al. did not make potatoes a free pass. It showed why a single-food GI table is a poor referee for a real meal.

GI still has a place. In people with diabetes, randomized trials pooled in a major meta-analysis found that lower-GI or lower-glycemic-load (GL) dietary patterns reduced HbA1c by 0.31 percentage points. The useful conclusion is narrower, and more practical, than "high-GI foods are bad": build a lower-GL diet pattern, but do not assume a food's isolated GI predicts your response to a mixed dinner.

Potatoes beat rice when the meal gets realistic

GI is measured by giving a fasted person a defined amount of available carbohydrate from one food, then comparing the blood-glucose response with a reference food. That protocol can compare foods under standardized conditions. It cannot reproduce dinner, where protein, fat, fiber, cooking method, portion size, timing, and the person eating the meal all affect glucose.

Amr et al. tested that gap directly in a randomized crossover study of 26 healthy, normal-weight adults without diabetes. Participants ate equivalent carbohydrate portions of instant mashed potatoes, baked French fries, or white rice as meals paired with either animal protein or plant protein. The animal-protein meals contained 26 g of protein and the plant-protein meals contained 10.4 g. Instant mashed potatoes had an isolated GI of about 84, baked fries about 77, and white rice about 73. By the usual GI ranking, rice should have produced the lower glucose response. Instead, both potato meals produced lower postprandial blood glucose than the rice meals.

The finding matters because it reverses the prediction made by the isolated GI values. It does not prove that every potato meal will outperform every rice meal, nor does it identify a winner between animal and plant protein. The protein conditions were not matched for protein amount, so the study cannot separate protein source from protein quantity. Its participants also had normal fasting glucose and did not have diabetes.

Still, the result did not remain confined to healthy volunteers. A crossover study by Devlin et al. in people with type 2 diabetes found lower nocturnal blood glucose after a potato-based mixed evening meal than after a rice-based equivalent. The published information available here does not provide the sample size, glucose area-under-the-curve values, or statistical details, so the effect should not be inflated beyond its direction. But it is a useful clinical replication of the central point: isolated GI rankings did not predict the response to a mixed meal.

Cooking changes the potato before the meal does

"Potato" is not a single glycemic exposure. In the Amr study, potato preparations spanned GI values of roughly 49 for potatoes refrigerated overnight to roughly 84 for instant mashed potatoes. Cooling cooked starch promotes retrogradation, which increases resistant starch; instant processing also disrupts the food matrix. Those changes help explain why preparations can differ so much before protein or other meal components enter the picture.

The range comes from the preparations assessed in that study, not every potato variety and recipe ever made. It does, however, support a sensible kitchen choice: if post-meal glucose is a priority, cooked potatoes cooled overnight are a better-supported option than instant mashed potatoes. Reheating may fit into that strategy, but this evidence does not quantify the glycemic effect of reheating.

The broader lesson extends beyond potatoes. A systematic review of randomized trials of reformulated breads found improved glycemic control when regular breads were changed to include features such as more fiber and less refined grain. And in randomized crossover studies of healthy adults, Chamoun et al. replaced 25 g of available carbohydrate from high-GI ingredients with cooked lentils in muffins, chili, and soup. Green lentils significantly lowered glucose incremental area under the curve versus the chili and soup controls; the muffin comparison did not reach statistical significance. A food matrix can change the response without changing the meal into a joyless spreadsheet.

GI works better as a dietary-pattern measure

None of this means GI is useless. Chiavaroli et al.'s 2021 BMJ systematic review and meta-analysis of randomized controlled trials, commissioned by the European Association for the Study of Diabetes Diabetes and Nutrition Study Group, found that low-GI/GL dietary patterns reduced HbA1c by 0.31 percentage points in people with diabetes. The reduction was statistically significant. HbA1c is a long-term glycemic marker, not a direct measure of complications or mortality, and the trials differed in comparator diets, follow-up duration, and feeding control. The meta-analysis therefore supports a modest, diet-achievable improvement rather than a precise promise for any one person.

The pattern evidence also avoids pretending that low GI is the only workable diet framework. A 2023 network meta-analysis by Jing et al. found that several dietary approaches emphasizing carbohydrate quality, including low-GI patterns, improved glycemic markers in type 2 diabetes; no single approach overwhelmed the others. A 2025 meta-analysis of randomized trials of Mediterranean diets likewise found significant improvements in glycemic control, including HbA1c, in adults with type 2 diabetes. Legumes, vegetables, whole grains, and olive oil can lower dietary GL without requiring anyone to treat a baked potato as contraband.

Your glucose response has a personal component

Population averages have limits. Popp et al.'s randomized clinical trial compared a personalized diet intended to reduce postprandial glycemic response with a low-fat diet in adults with abnormal glucose metabolism and obesity, and it identifies substantial person-to-person variation in responses to the same foods as a challenge for universal GI advice. That observation does not establish that personalized nutrition will beat standard care for every outcome. It does explain why a food ranking can fit one person poorly.

A 2025 observational analysis by Calvo-Malvar et al. found that age, sex, BMI, and meal timing independently modified glycemic response to a given meal GL in a general-population setting. These are associations, not proof that changing one factor will cause a specified glucose change. They also sit outside conventional GI testing.

Gut biology may account for some of the variation. A meta-analysis and umbrella review by Xu et al. found that higher intake of microbiota-accessible carbohydrates improved cardiometabolic risk factors in type 2 diabetes. Resistant starch is one such fermentable carbohydrate and can feed short-chain-fatty-acid-producing microbes. Separately, Skantze et al. linked different responses to a standardized mixed meal with type 2 diabetes risk factors and gut microbiota composition. These findings provide a plausible route by which cooled-potato resistant starch could affect people differently; they do not prove that a person's microbiome will predict their response to a particular potato dish.

What to do with this at dinner

• Use GI tables as rough population-level information, not meal-by-meal verdicts.
• Pair carbohydrate foods with protein, fiber-rich vegetables, legumes, and other minimally processed foods.
• Choose less processed potato preparations; cooled cooked potatoes are reasonable when glucose management matters.
• For diabetes, focus on the whole dietary pattern. Lower-GI/GL patterns can modestly improve HbA1c, but they are one evidence-based option among several.
• If you use a glucose monitor, look for repeatable responses to your own usual meals rather than drawing conclusions from one dramatic trace.

Your potato with protein can fit a glucose-conscious meal. The stronger target is the pattern you eat repeatedly: its fiber, processing level, portions, and overall carbohydrate quality.