Resistant Starch: From Cooled Potatoes to the Insulin Trials

Cooling cooked potatoes or rice and eating them later is one of the few kitchen “hacks” with a real mechanism and human trials behind it. It is also oversold. The resistant starch that forms in the fridge is real, but the amounts are small, the glucose effect is modest, and whether it feeds your gut bacteria depends on which bacteria you already have.

What resistant starch is and why cooling matters

Freshly cooked starch digests quickly. As it cools, part of it re-forms into a tighter structure that digestive enzymes struggle to break down. Food scientists call this retrogradation, and the rice trial below starts from that premise: the fridge is where retrograded, resistant starch forms.

Starch that escapes digestion reaches the colon, where bacteria ferment it into short-chain fatty acids, including butyrate. So part of the starch behaves like a fermentable fibre rather than a sugar. The fibre gap post covers what those fatty acids do once they are made, and butyrate is the compound the gut barrier post keeps returning to.

Resistant starch also comes as powders unrelated to leftovers, such as raw potato starch and high-amylose maize starch; the supplement trials below used purified starch, not food.

Cooled rice: what 24 hours in the fridge does

The cleanest food study is Sonia 2015, a randomised single-blind crossover in 15 healthy adults. White rice was tested fresh, cooled 10 hours at room temperature, or cooled 24 hours at 4°C then reheated: 0.64, 1.30 and 1.65 g of resistant starch per 100 g. Refrigerating and reheating gave roughly two and a half times the resistant starch of fresh rice.

In the feeding test the refrigerated-and-reheated rice produced a lower glucose response than fresh rice, an area under the curve of 125 versus 152 mmol·min/L, about 18 percent less and only just significant (p = 0.047). Two things to hold on to: reheating did not erase the gain over fresh rice, and the absolute amount is still small. At 1.65 g per 100 g, a 150 g serving carries about 2.5 g of resistant starch, against roughly 1 g fresh.

Cooled potatoes and what reheating does

Raatz 2016 cooked three common varieties by baking or boiling and served them hot at 65°C, chilled at 4°C, or chilled for six days and reheated to 65°C. Cooking method and serving temperature changed the resistant starch content; variety did not. Baking beat boiling, and chilled beat both hot and reheated. With potatoes, reheating gave back part of what chilling created, so cold keeps the most; the rice trial never measured cold, unreheated rice, so it cannot say whether rice behaves the same way.

Leeman 2005 ran a meal test in 13 healthy adults. Storing boiled potatoes 24 hours at 8°C raised resistant starch from 3.3 to 5.2 percent of the starch. Against a white-bread reference, cold storage alone cut the insulin response by 28 percent; cold potatoes with a vinaigrette (28 g white vinegar, 8 g olive oil) cut the glycaemic response by 43 percent and the insulin response by 31 percent versus freshly boiled. Cooling and vinegar worked separately, which makes potato salad the combination the trial favoured.

Patterson 2019 tested it in 30 women with elevated fasting glucose and insulin (mean BMI 32.8), using 250 g portions in a randomised crossover. Baked-then-chilled potatoes, against boiled, lowered glucose by 4.8 and 9.2 percent at 15 and 30 minutes, insulin by 25.8 and 22.6 percent, and the gut hormone GIP by 41.1 and 37.6 percent. Over the whole curve, insulin and GIP were significantly lower while glucose was not. The pattern is a small glucose difference and a much larger insulin difference.

The insulin trials

The insulin-sensitivity claim rests on a handful of small supplement trials.

Robertson 2005 gave 10 healthy adults 30 g per day of resistant starch or placebo for four weeks. Clamp-measured insulin sensitivity was higher on resistant starch (9.7 versus 8.5 units, p = 0.03) and meal-test insulin sensitivity was 33 percent higher (p = 0.05). Insulin-adjusted glucose clearance by forearm muscle was 44 percent higher, and short-chain fatty acids rose during the meal, which is the proposed pathway. Ten people, four weeks.

Maki 2012 complicates the story. Thirty-three adults (mean BMI 30.6, waist 105 cm) took 0, 15 or 30 g per day of high-amylose maize resistant starch in random order for four-week periods, double-blind. In the 11 men the insulin sensitivity index was 4.66 on placebo and 6.90 and 7.13 (×10⁻⁵) on the two doses; in the 22 women no dose differed from placebo. The sex interaction was significant (p = 0.033), and two authors were employees of the starch manufacturer.

Johnston 2010 ran the longest trial: 20 insulin-resistant adults with metabolic syndrome took 40 g per day or placebo for 12 weeks. Clamp insulin sensitivity improved versus placebo (p = 0.023); body weight, liver fat, visceral fat, vascular function and inflammatory markers did not change.

The pooled data are cooler. Xiong 2021 combined 19 randomised trials and found fasting glucose 0.09 mmol/L lower on resistant starch than on digestible starch (95% CI −0.13 to −0.04), more at doses above 28 g per day (−0.16) and beyond eight weeks (−0.12); HOMA-IR fell (effect size −0.33), but fasting insulin and four glucose-tolerance-test endpoints did not change significantly. Snelson 2019 pooled 22 trials of type 2 resistant starch (670 participants) and found no effect on satiety, fasting glucose, HbA1c or insulin resistance; its only positive findings, on triglycerides and weight, were driven by a few studies contradicting the majority, in trials of one to 12 weeks with 10 to 60 people each. A 0.09 mmol/L change in fasting glucose is about 1.6 mg/dL.

Why the gut response is person-dependent

The fermentation side has a clearer explanation: not everyone carries the bacteria that break resistant starch down.

Walker 2011 fed 14 overweight men four controlled diets in sequence over 10 weeks. On the resistant-starch diet, ruminococci related to Ruminococcus bromii, a known starch degrader, averaged 17 percent of bacteria detected, against 3.8 percent on a non-starch-fibre diet. Two volunteers carried under 1 percent of these bacteria, and in them more than 60 percent of the resistant starch passed through unfermented, against under 4 percent in the other 12.

Venkataraman 2016 had 20 healthy young adults ramp up over three days to 48 g per day of raw potato starch (about half of it resistant starch) and hold that dose for a week. Faecal butyrate rose on average from 8 to 12 mmol/kg, but the average hid three groups: 11 “enhanced” responders whose butyrate rose 67 percent, three already high, and six with low butyrate that did not budge. Starch degraders such as Bifidobacterium adolescentis and R. bromii reached around 9 percent of the community in responders and stayed near 1.5 percent in the low group.

Baxter 2019 scaled this to 174 healthy young adults comparing roughly 28 to 34 g per day of resistant potato starch, 20 to 24 g of resistant maize starch, 20 g of inulin and a digestible-starch control over three weeks, with a half-dose run-in of four to seven days. Potato starch drove the biggest rise in short-chain fatty acids, butyrate included; maize starch and inulin changed the community without significantly raising butyrate. Even on potato starch, butyrate rose in 63 percent of people and stayed flat or fell in 37 percent; the gainers were those in whom R. bromii or Clostridium chartatabidum rose alongside the butyrate producer Eubacterium rectale.

So “resistant starch feeds your good bacteria” held, by the butyrate measure, for roughly two in three people in the largest trial. In a fibre and fermented-food routine, it is one input among several.

Food versus supplement: how much is realistic

Food and supplement numbers live in different ranges. The trials that moved insulin sensitivity used 30 to 40 g per day; the butyrate studies used 40 to 48 g of raw potato starch powder, roughly 24 to 34 g of actual resistant starch. Cooled-and-reheated rice at 1.65 g per 100 g would take close to two kilograms a day to match 30 g. Cooled potatoes do better, but at Leeman’s 5.2 percent of the starch most of a cold potato is still digestible starch.

The meal studies show what a single cooled portion does to that meal’s glucose and insulin: a modest reduction, larger for insulin than glucose. “Cooled potato salad improves insulin sensitivity” borrows the supplement results for a dose the food cannot deliver.

Two more things: the microbiome trials ramped up gradually, and responses split by sex in one trial and by microbiome in three. Anyone with a diagnosed metabolic condition, on glucose-lowering medication, or pregnant should talk to a clinician before adding 30 g or more of any fermentable starch to their day.

Bottom line

Cooling cooked starch raises its resistant starch content; reheated rice still beat fresh rice, while reheated potatoes lost part of what chilling added. Meal tests show a lower glucose response and a larger insulin drop from cooled rice and potatoes, largest when cold potatoes are dressed with vinegar. The insulin-sensitivity improvements come from supplement trials at 30 to 40 g per day, appeared in men but not women in one trial, and shrink to small changes in the meta-analyses. The gut effect is real, potato starch beat maize starch and inulin for butyrate, and about a third of people did not respond, apparently because they lack the bacteria that ferment it.