Semolina: the shape it is cooked into does the work

Semolina is durum wheat milled coarse, and what it is made into moves blood sugar more than the grain does. In 30 healthy Chinese and Indian adults aged 21 to 40, semolina spaghetti came out at a glycemic index of 53 and wholegrain spaghetti at 54, against 74 for wheat mee pok noodles and 80 for white rice measured in the same people. The difference shows up in the total load too. Over 120 minutes the glucose area under the curve after either spaghetti was lower than after the noodles or the rice, with no gap between the refined and the wholegrain pasta. These were single test meals in 30 normal-weight healthy adults, and the product was dried industrial spaghetti rather than semolina cooked as porridge. What is milled in alongside it can move the number further. In 9 healthy adults, spaghetti carrying 10 percent of one barley beta-glucan concentrate had a glycemic index of 29 against 64 for the control, while a second concentrate at the same dose changed nothing.

The numbers, per 100 g

How much is actually in it

• US law defines semolina as cleaned durum wheat ground and bolted to a set fineness and freed from bran coat, with ash no more than 0.92 percent and moisture no more than 15 percent, which is what separates it from ordinary wheat flour. — not applicable, this is a composition rule for product sold in the United States Evidence E source
  Passes a No. 20 sieve with no more than 3 percent through a No. 100 sieve, ash no more than 0.92 percent on a moisture-free basis, moisture no more than 15 percent
• One number for semolina hides a wide spread. Across eight USDA Foundation samples of fine semolina flour, iron ran from 0.605 to 4.64 mg per 100 g and niacin from 1.91 to 9.08 mg, while folate stayed within a third of its median. — not applicable, eight analysed samples of fine semolina flour in USDA Foundation Foods (n = 8 samples) Evidence D source
  Iron spread 166 percent, niacin 128 percent, molybdenum 116 percent, manganese 87 percent, magnesium 72 percent, folate 33 percent

What your body absorbs

• Slower starch digestion in a test tube did not carry over into people: semolina pudding from a high resistant starch durum mutant digested far slower in vitro at C90 33.29 percent against 47.38 percent, yet showed no glycemic index difference in 10 healthy adults. — in vitro digestion of retrograded semolina, then 10 healthy participants eating about 50 g of total starch as pudding with 2 hour capillary glucose sampling (n = 10 in the human arm) Evidence D source
  In vitro C90 33.29 percent against 47.38 percent, no evidence of a difference in glycemic index in vivo

What cooking and storage change

• How semolina is cooked changes the blood sugar answer: among 30 people with type 2 diabetes given 50 g carbohydrate portions, only the steam-cooked semolina meal and the semolina with Bengal gram dhal came out significantly lower than glucose at 1 hour. — 30 patients with non-insulin-dependent diabetes, five conventional Indian meals of 50 g carbohydrate each, semolina cooked by two different methods (n = 30) Evidence B source
  No significant differences in mean GI across meals except roasted semolina and semolina with black gram dhal, and only steam-cooked semolina and semolina with Bengal gram dhal fell below the glucose reference at 1 hour

What it does to your health

• Spaghetti made from semolina behaves as a low glycemic index food at GI 53, against 74 for wheat mee pok noodles and 80 for white rice measured in the same 30 people. — 30 healthy Chinese and Indian adults aged 21 to 40 with BMI 18.5 to 25, seven test sessions with 3 hours of blood sampling each (n = 30) Evidence B source
  GI 53 for semolina spaghetti and 54 for wholegrain spaghetti against 74 for noodles and 80 for rice (p < 0.005), with lower insulin response than rice (p < 0.05)
• The same trial found the 120 minute glucose area under the curve significantly lower after semolina spaghetti than after noodles or rice, so the gap is in total glucose load and not only in the index. — 30 healthy Chinese and Indian adults, glucose and insulin measured over 180 minutes after each of seven test meals (n = 30) Evidence B source
  Glucose iAUC over 120 min lower than noodles (p < 0.005) and rice (p < 0.005), no difference between semolina and wholegrain spaghetti
• What is milled into the semolina decides the spike: 10 percent of one barley beta-glucan concentrate cut the glycemic index of semolina spaghetti from 64 to 29 in nine healthy adults, while the other concentrate at the same dose changed nothing. — 9 healthy subjects, spaghetti samples with 0, 2, 4, 6, 8 and 10 percent beta-glucan from two barley concentrates, 50 g available carbohydrate per portion (n = 9) Evidence B source
  GI 29 for 10 percent Barley Balance spaghetti against 64 for control (54 percent lower, P <= 0.02), IAUC 52 percent lower (P <= 0.017), Glucagel spaghetti not different from control

Safety, limits and interactions

• Semolina is durum wheat, so for wheat allergy it is the allergen rather than a mild alternative: 12 patients with wheat-induced anaphylaxis went through an oral immunotherapy protocol that started with semolina flour and ended with spaghetti under medical supervision, and they reached tolerance of 50 g of wheat. That is a hospital protocol, not something to repeat at home. — 12 patients, children and preschool children among them, with IgE-mediated wheat-induced anaphylaxis, followed with IgE assays before and after and 18 months later (n = 12) Evidence C source
  Tolerance of 50 g of wheat reached, total IgE fell after up-dosing and at follow-up (p < 0.01), wheat-specific IgE rose after up-dosing and fell at follow-up (p < 0.001)

What people believe that the data does not support

• Refined does not automatically mean a bigger blood sugar spike. With grain source, food form, particle size and viscosity matched, semolina porridge gave a lower 0 to 120 minute glucose area than whole wheat flour porridge in 16 adults, a mean difference of -966 mg min/dL, in a trial the authors read as showing little-to-no difference overall. — 16 healthy participants, six medium-viscosity porridges from the same grain source and mill, continuous glucose monitoring plus breath test for gastric emptying (n = 16) Evidence B source
  Semolina mean difference -966 mg min/dL (95% CI -1775 to -156, P = 0.02) and cracked wheat -721 mg min/dL (95% CI -1426 to -16, P = 0.04) against whole wheat flour porridge, with Bayes factors supporting a null effect overall

The bottom line

The belief worth checking is that refined always spikes harder. Sixteen healthy adults ate six porridges made from the same grain and the same mill, matched for particle size and viscosity, and the semolina porridge gave a lower 0 to 120 minute glucose area than the whole wheat flour porridge, a mean difference of -966 mg min/dL (95 percent CI -1775 to -156, P = 0.02). The authors read their own trial as showing little-to-no difference overall, with Bayes factors supporting a null. The honest version is that matching the form took the advantage away from the wholegrain rather than handing one to the refined flour. Cooking is the lever a reader actually holds. Among 30 people with type 2 diabetes eating meals of 50 g of carbohydrate each, only the steam-cooked semolina and the semolina with Bengal gram dhal produced a significantly lower blood glucose response than a 50 g glucose load at one hour. That trial is from 1992, the meals mixed semolina with pulses so the cooking method and the pulse cannot be separated, and everyone in it had diabetes. A slower test tube did not mean a flatter curve. Semolina from a high resistant starch durum mutant digested far slower in a laboratory assay, at C90 33.29 percent against 47.38 percent, and in a 10 person pilot the pudding made from it showed no glycemic index difference from the control. Ten people finding nothing is weak evidence rather than proof of no effect. Two records fix what the word means and how far it stretches. US law defines semolina as cleaned durum wheat ground and bolted fine enough to pass a No. 20 sieve, freed from bran coat to an ash content of no more than 0.92 percent on a moisture-free basis, with moisture capped at 15 percent. That is a labelling floor rather than a nutrient measurement. Across eight analysed USDA Foundation samples of fine semolina flour, iron ran from 0.605 to 4.64 mg per 100 g and niacin from 1.91 to 9.08 mg, while folate stayed inside a third of its median. Those are analysed samples rather than a designed comparison, and the spread follows wheat lot, soil and mill. Enrichment adds a second source of spread, which is why the composition table below carries an enriched and an unenriched semolina side by side. The caution is plain. Semolina is durum wheat, so for anyone with wheat allergy it is the allergen. Twelve patients with IgE-mediated wheat-induced anaphylaxis, children among them, went through an oral immunotherapy protocol that began with semolina flour and ended with spaghetti, and they reached tolerance of 50 g of wheat. That was twelve patients with no control group, run under medical supervision, and it belongs here as evidence that semolina carries the wheat allergen rather than as anything to try. We have no card on pregnancy, on kidney disease or on drug interactions for semolina.

Sources

1. semo-daily-r4-08 · Code of Federal Regulations, standard of identity
   Semolina is the food prepared by grinding and bolting cleaned durum wheat to such fineness that, when tested by the method prescribed in § 137.300(b)(2), it passes through a No. 20 sieve, but not more than 3 percent passes through a No. 100 sieve. It is freed from bran coat, or bran coat and germ, to such extent that the percent of ash therein, calculated to a moisture-free basis, is not more than 0.92 percent. Its moisture content is not more than 15 percent.
   21 CFR 137.320, Semolina · checked 2026-09-24
2. semo-daily-r4-09 · food composition dataset, Foundation Foods sample ranges
   USDA FoodData Central, Foundation Foods, release 2026-04-30 | Flour, semolina, fine | 8 samples | iron 2.431 mg per 100 g, range 0.605-4.64, spread 166% | niacin 5.599 mg, range 1.91-9.08, spread 128% | molybdenum 47.91 ug, range 24.5-80.2, spread 116% | magnesium 46.28 mg, range 24.3-57.8, spread 72% | folate 163.6 ug, range 131.0-185.0, spread 33%
   USDA FoodData Central, Foundation Foods, release 2026-04-30 · checked 2026-09-24
3. semo-daily-r4-06 · in vitro starch digestibility assay with a small human pilot
   We found no evidence of difference in GI between sbeIIa/b-AB pudding and the WT control, but the starch digestibility was significantly lower in sbeIIa/b-AB pudding compared to the WT control in vitro (C90 = 33.29% and 47.38%, respectively).
   Food Funct, 2020 · checked 2026-09-24
4. semo-daily-r4-03 · randomised controlled trial in patients with type 2 diabetes
   Compared with the blood glucose response for a 50 g glucose load, only meals based on steam-cooked semolina and semolina-Bengal gram dhal elicited a significantly lower response at 1 h postprandially, and only meals based on semolina-black gram dhal at 2 h postprandially.
   Br J Nutr, 1992 · checked 2026-09-24
5. semo-daily-r4-01 · randomised crossover trial in healthy adults
   The current evaluation showed that semolina spaghetti and wholegrain spaghetti can be classified as low GI products, with a GI of 53 and 54, respectively, significantly lower than wheat based mee pok noodles (74) and rice (80) (p < 0.005).
   Nutrients, 2021 · checked 2026-09-24
6. semo-daily-r4-02 · randomised crossover trial in healthy adults
   The glucose iAUC (120 min) was significantly lower for both semolina and wholegrain spaghetti as compared to noodles (p < 0.005) and rice (p < 0.005).
   Nutrients, 2021 · checked 2026-09-24
7. semo-daily-r4-05 · randomised controlled trial in healthy adults
   In particular, the GI of 10% BB spaghetti was 54% lower (P ≤0.02; GI = 29) than that of the control (GI = 64).
   Nutrition, 2011 · checked 2026-09-24
8. semo-daily-r4-07 · non-randomised clinical trial in patients with wheat anaphylaxis
   12 patients underwent OIT protocol that comprised of two phases: the first with semolina flour and the second with spaghetti...Patients successfully tolerated 50 g of wheat.
   Immunotherapy, 2017 · checked 2026-09-24
9. semo-daily-r4-04 · randomised crossover trial in healthy adults
   Little-to-no differences were observed in glycemic responses, with lower incremental AUC between 0 and 120 min glycemic responses only for semolina [mean difference (MD): -966 mg min/dL; 95% CI: -1775, -156 mg min/dL; P = 0.02) and cracked wheat (MD: -721 mg min/dL; 95% CI: -1426, -16 mg min/dL; P = 0.04) than for whole wheat flour porridge.
   Am J Clin Nutr, 2022 · checked 2026-09-24