Is a vegan diet good for diabetics?

Type 2 diabetes and prediabetes are driven largely by insulin resistance, and diet is one of the few levers a person can adjust without a new prescription. The evidence on whether a plant-based diet specifically helps is real and moderately sized - but it comes with two catches before switching: the diet has to be built mainly from whole foods, not the vegan version of fries and cookies, and a diagnosed diabetic starting one needs to loop in their doctor, because glucose-lowering medication doses tend to need adjusting within weeks, not months.

What the trials show

The foundational trial is Neal Barnard's 2006 randomized study: 99 people with type 2 diabetes assigned to a low-fat plant-based diet or the standard ADA-recommended diabetic diet for 22 weeks. Among participants who didn't change medication during the trial, A1c fell 1.23 points on the plant-based diet against 0.38 on the ADA diet (P=0.01), alongside more weight loss (6.5kg vs 3.1kg) and a bigger LDL drop (21.2% vs 10.7%). Followed to 74 weeks, the plant-based group's A1c had dropped further, −0.34 points versus −0.14, but that longer-term gap wasn't statistically reliable on its own, meaning it could have been due to chance. It only held up once researchers excluded people whose diabetes medication had already been changed during the trial - and that adjustment itself makes the number hard to fully trust either way.

Guest et al.'s 2024 GRADE-rated meta-analysis of 7 trials found a real, moderate-certainty A1c improvement and a drop in BMI - but fasting glucose and LDL cholesterol didn't move, so the same trials that back the headline A1c number don't back a blood-sugar or cholesterol claim beyond it. An earlier meta-analysis of 6 trials found nearly the same A1c effect size independently, funded by Nestlé Nutrition Council Japan rather than a vegan-advocacy group - a useful check, since several of the largest individual trials, Barnard's included, along with a run of more recent ones by researcher Hana Kahleova, were funded by the Physicians Committee for Responsible Medicine, an advocacy nonprofit that promotes plant-based eating. That doesn't invalidate the trials; funding source and independent replication are two different questions, and this effect size has been replicated by independently funded work. A 0.4-point A1c drop is roughly half the effect of adding metformin, the first-line diabetes drug, which lowers A1c by about 0.9 points on its own.

It has to be mainly whole food, not just plant-based onion rings

A plant-forward diet isn't automatically a healthy one. A 2016 analysis of three long-running US cohorts (over 4.1 million person-years, 16,162 new type 2 diabetes cases) split "plant-based" eating into two versions scored the same way: a healthful pattern built on vegetables, whole grains, fruit, legumes and nuts, and an unhealthful one built on refined grains, fried potatoes, and sugar-sweetened drinks, both still technically low in animal food. People eating the healthful pattern had 45% lower diabetes risk at the extremes of the comparison. People eating the unhealthful pattern had 16% higher risk. Both counted as "plant-based" by the same scoring method. The Academy of Nutrition and Dietetics' 2025 position paper reaches the same conclusion: restricting animal products alone doesn't reduce cardiometabolic risk, and the protective effect depends on actually eating the foods that carry it, soy, legumes and viscous fiber by name.

A 2025 secondary analysis of a 16-week randomized trial looked at whether the degree of food processing blunted the weight-loss benefit of a plant-based diet, and didn't find that it did - processed plant foods weren't associated with less weight loss in that trial. It's a single trial, a secondary analysis rather than a pre-registered one, and it's about weight rather than A1c specifically, so it doesn't overturn the cohort finding above.

Why it works

In a 16-week randomized trial, participants on a low-fat plant-based diet lost an average 6.4kg, against 0.5kg on their usual diet. Fat stored inside the liver and muscle cells, a known driver of insulin resistance, dropped 34% and 10% respectively, alongside a 0.9-point rise on the trial's standard insulin-sensitivity index. A related analysis of the same trial's gut-microbiome data found that increases in one bacterial species, Faecalibacterium prausnitzii, correlated with the size of a person's insulin-sensitivity gain. The likely levers behind all of it are more fiber, less saturated fat, and a lower calorie density that makes the weight loss easier to sustain.

The catch: medication doses move fast

In a 12-week randomized trial of adults with type 1 diabetes, the plant-based group cut their total daily insulin dose by 28%, about 12 units a day, while the control group's dose didn't move. In a 24-week whole-food plant-based trial for type 2 diabetes, 63% of medicated participants reduced their glucose-lowering medication against 24% in the control group, and 23% of those who started with an A1c under 9% reached remission.

A diabetic who starts eating this way without adjusting medication or insulin dose in step is running toward hypoglycemia. Every trial cited here had participants under medical supervision with dose adjustments built into the protocol. The Academy's own 2025 review rates the medication-reduction evidence itself as very low quality, even though the individual trials above show large effects - a doctor's involvement decides the actual dose change, not a formula.

What "well planned" requires

B12 has to come from a fortified food or a supplement; no plant food reliably provides it, and on an unsupplemented plant-based diet, a deficiency can take years to show up, long after the diet switch that caused it. Bean- and legume-heavy meals, the backbone of most of the trial diets above, carry more carbohydrate per gram of protein than meat or dairy do, so carb-counting has to shift even though the fiber in those same foods slows the glucose response enough to help offset it. Iron absorbs less efficiently from plant sources than from meat, the same 5-12% versus 14-18% gap that applies to any vegan diet, and anemia complicates diabetes management further, so the same pairing-and-timing planning applies here too. And for diabetics managing kidney involvement specifically, the American Diabetes Association's own guidance flags that high plant-protein foods, beans, legumes, nuts and tofu, run higher in potassium and phosphorus, and recommends working with a dietitian.

If you're already vegan and diabetic

Check that meals are mainly built on legumes, whole grains, vegetables, fruit and nuts rather than processed vegan substitutes, since that distinction carries most of the effect size above. Confirm B12, and potassium or phosphorus if kidney involvement is part of your case, with your care team rather than assuming the diet handles it on its own.

If you're a diabetic considering the switch

Loop in whoever manages your medication. Bring the trial numbers to that conversation: a 28% average insulin-dose reduction in the type 1 trial, a 63% medication-reduction rate in the type 2 one, so the adjustment gets planned rather than caught after the fact.