Can Intermittent Fasting Put Type 2 Diabetes Into Remission?
Can intermittent fasting put type 2 diabetes into remission? What the trials found, what remission really means, and why diabetes medicines need a plan first.
Headlines about fasting “reversing” diabetes tend to arrive with one striking percentage: 80% of people in one trial, nearly half in another. If you have type 2 diabetes, the obvious question is whether skipping meals could get you off your tablets too.
Some people do reach remission after a structured fasting-style diet. But those trials tested supervised low-calorie programs, and they didn’t measure the same thing.
The short version
Some people with type 2 diabetes reach remission after supervised, low-calorie diet programs, including one small trial of an intermittent program. Remission means HbA1c below 6.5% after at least three months without glucose-lowering medicines; a lifestyle intervention also needs time to settle. A widely reported 80% result from a different trial measured low HbA1c at 16 weeks, not remission. These studies changed calories, food and weight as well as eating patterns, so they don’t establish that an ordinary 16:8 window produces remission. Ask your prescriber before fasting, especially if you use insulin or a sulfonylurea: eating less can cause dangerous low blood sugar.
What “remission” actually means
In 2021, an international panel convened by the American Diabetes Association and other diabetes organizations agreed on a working definition. Remission means an HbA1c below 6.5% (48 mmol/mol) measured at least three months after stopping all glucose-lowering medication (Riddle et al., Diabetes Care 2021).
HbA1c reflects your average blood glucose over roughly the past three months. That is why the timing rules matter. The panel says a lifestyle change can take up to six months to settle, so it advises waiting at least six months after starting one before judging the response. A single good reading a few weeks into a new diet doesn’t qualify. Nor does one good result while treatment continues establish medication-free remission. Some medicines may remain appropriate for heart, kidney or other benefits even when glucose improves. The label is not a reason to stop useful treatment.
The panel deliberately chose “remission” over “reversal” or “cure”. Glucose can rise again, and the damage that years of high glucose can leave behind doesn’t vanish. People in remission are still advised to keep up eye screening, kidney tests, foot checks and blood pressure monitoring, with HbA1c checked at least once a year.
The trial that actually measured remission
Yang and colleagues (2023) randomized 72 Chinese adults with type 2 diabetes to an intermittent calorie-restricted program or usual dietary care. Participants were taking glucose-lowering medicines; mean BMI was about 24.
The intervention repeated six 15-day cycles: five days of supplied food kits, about 840 calories daily, followed by ten days of unrestricted intake. Both groups received diabetes dietary guidance, and physicians managed medicines. This was a specific three-month food program, not a trial of skipping breakfast or a standard 16:8 window.
After another three months, remission was reported in 17/36 participants versus 1/36 controls. At the 12-month follow-up it was 16/36 versus 0/36. The reported odds ratio at the first follow-up had a very wide 95% confidence interval, 2.39–121.07: the size of the benefit is uncertain.
Nine people withdrew. Remission used all 72 randomized participants as denominators; continuous outcomes used 63. The intervention group’s roughly 6 kg mean weight loss accompanied the diet, so meal timing’s separate effect is unknown. The small, open-label trial was registered after recruitment began. It reported no serious adverse events, which does not establish that hypoglycemia never occurred.
The bigger trial with a smaller claim
The 80% figure comes from the EARLY trial, by Guo and colleagues (2024) in JAMA Network Open. It randomized 405 Chinese adults with overweight or obesity who had been diagnosed within the past year and had not taken diabetes medicine for at least three months.
One group followed a 5:2 plan: on two nonconsecutive days a week, a meal-replacement product replaced all three meals, at 500 calories for women and 600 for men. On the other five days, people ate their own breakfast and lunch, and a second meal-replacement product replaced dinner. The other groups took metformin or empagliflozin, two common diabetes drugs. All three groups received dietary and exercise guidance and general diabetes education every four weeks; the comparison was not diet support versus pills alone.
After 16 weeks, HbA1c fell by 1.9 percentage points in the 5:2 group, compared with 1.6 in the metformin group and 1.5 in the empagliflozin group. The adjusted differences were 0.3 and 0.4 percentage points, respectively. The paper reports inconsistent confidence intervals across its abstract, results and figure legend, so we do not reproduce a disputed interval or treat those point estimates as exact. Average weight loss was 9.7 kg in the 5:2 group, against about 5.5 to 5.8 kg with the drugs.
At week 16, 108 of 135 people (80.0%) in the 5:2 group had HbA1c below 6.5%. That proportion describes an HbA1c threshold, not an established remission rate. The authors themselves note the intervention was shorter than the six months the definition requires. Eight weeks later, 72 of the 94 people in the 5:2 group who had a measurement (76.6%) still had HbA1c below 6.5%; the remaining 41 had no week-24 value reported.
Two trials · share of each group · one 0–100% scale
A low HbA1c reading is not the same as remission
Both studies counted people with HbA1c below 6.5%. Only the first assessed remission after the medication-free interval.
Remission was tested
Yang 2023 · 72 adults already on diabetes drugs · China
Counted: HbA1c below 6.5% at least 3 months after stopping all diabetes medication
-
Fasting-day kits
5 low-calorie days in every 15, for 3 months
-
Usual diabetes diet
National dietary guidelines
Remission was not tested
EARLY 2024 · 405 newly diagnosed adults, no drugs at entry · China
Counted: HbA1c below 6.5% at week 16. The drug groups were still taking their drug
-
5:2 meal-replacement plan
2 days of 500–600 kcal, plus a replacement dinner on other days
-
Metformin
Medication + lifestyle guidance
-
Empagliflozin
Medication + lifestyle guidance
The trial was funded by the company that makes the meal-replacement products. The paper says the funder had no role in the design or analysis. Participants and investigators knew who was getting which treatment. Only 332 of 405 completed 16 weeks; the main analysis retained all 405, using multiple imputation for missing outcomes. The plan also changed more than two days a week: dinner was replaced every day.
Why weight loss keeps showing up
Neither trial isolates fasting from weight loss. Changing the food supplied, calorie intake, clinical support and medicine use together makes it difficult to identify what produced the benefit.
The UK’s DiRECT trial (Lean et al., 2018) tested a supervised formula diet rather than intermittent fasting. Participants replaced food with 825–853 calories daily for three to five months, then gradually reintroduced food with maintenance support. Clinicians withdrew oral diabetes and blood-pressure drugs at the start of the program and reintroduced them when needed. This was a treatment protocol, not advice to stop medicines yourself.
At 12 months, 68/149 intervention participants (46%) were in remission, versus 6/149 controls (4%). Its definition required at least two months off diabetes medicines, predating the later three-month consensus. Missing primary outcomes counted as failures. Participants had BMI 27–45, diabetes for less than six years and were not using insulin, so this result does not apply equally to everyone with type 2 diabetes.
Across 285 people with weight observations, remission was more common among those who lost more weight.
DiRECT trial · 12 months · 285 people grouped by weight change
Remission tracked how much weight came off
Share of people in remission (HbA1c below 6.5% after at least two months off all diabetes drugs), sorted by how much weight they had lost.
-
Gained weight
-
Lost 0–5 kg
-
Lost 5–10 kg
-
Lost 10–15 kg
-
Lost 15 kg or more
The diet was a supervised 825–853 kcal formula plan for 3 to 5 months, not intermittent fasting. Groups are formed by weight lost after the fact, so they show an association, not a tested dose.
That is an association across weight-loss categories, not a randomized test of losing a specified number of kilograms. It supports discussing weight management with a care team, but it cannot tell you what you personally need to lose or prove that meal timing is irrelevant.
Where 16:8 fits
Most people who say “intermittent fasting” mean a daily eating window, such as 16:8. The eating-window trial discussed here tested weight and glucose outcomes, not remission.
In a six-month trial in Chicago, Pavlou and colleagues (2023) assigned 75 adults with type 2 diabetes to eating between noon and 8 p.m., to daily calorie counting, or to a control group. Both diet groups lowered HbA1c by about 0.9 percentage points relative to control. The direct eating-window versus calorie-restriction difference was 0.04 points (95% CI −0.64 to 0.72). That interval leaves meaningful differences possible; it is not proof of equivalence. The eating-window group lost 3.6% of body weight relative to control. The trial did not measure remission. It was small, unblinded and powered primarily for weight, with self-reported intake; its secondary glucose comparisons need caution.
In Australia, Carter and colleagues (2018) compared a 5:2 plan with daily calorie restriction in 137 adults for 12 months. HbA1c fell by 0.3 points with 5:2 and 0.5 with daily restriction, a difference that met the trial’s prespecified HbA1c equivalence test: 0.2 points, with a 90% confidence interval of −0.2 to 0.5 within a ±0.5-point margin. This did not establish equivalence for weight loss or body composition. Only 97 of 137 participants completed the year; the analysis included everyone randomized under a missing-at-random assumption. Remission was not the endpoint.
These trials support possible improvements in glucose control under supervision. They do not establish remission from a daily window, and Carter’s 5:2 result cannot be transferred to 16:8. This is a targeted review of selected trials, not a systematic search for every remission study. For more on what windows do to blood sugar in people without diabetes, see our look at whether 16:8 lowers blood sugar.
The real risk: fasting on glucose-lowering drugs
Every trial above handled medication carefully. Insulin and sulfonylureas, such as gliclazide and glimepiride, can cause low blood sugar when meals are missed or delayed (NHS hypoglycemia guidance).
In the Carter trial, the medication protocol had to be tightened partway through, mainly because of hypoglycemia. In the first two weeks, 16 of 46 people on sulfonylureas or insulin (35%) had a low or high glucose event across both diet groups; 8 of 46 had hypoglycemia. The 35% figure is not a rate of low blood sugar alone. These people had a dietitian, an endocrinologist and a written plan.
The remission panel adds a less obvious caution. In people who already have more than minimal diabetic eye disease, a rapid fall in glucose after years of high readings can temporarily worsen the retina. That’s a reason to tell your eye-screening team before starting an aggressive plan, not a reason to stay at high glucose.
How much to trust this
The fasting-style evidence reviewed here includes one small remission trial and one larger, short glucose-control trial. Both supplied specific food products. Neither establishes that choosing a fasting schedule on your own reproduces the result.
The populations also differ. Yang included relatively lean adults, while EARLY enrolled people diagnosed within a year without recent diabetes medication. DiRECT excluded insulin users and used a different remission definition. Comparing their percentages as if they were a league table would hide those differences.
Remission can also end. Follow-up lasting around a year cannot settle what happens over five or ten years. Weight management may be a useful treatment goal, but a person’s response, suitable diet and medication needs require individual follow-up. Continuing medication is not a failure; improved glucose control can matter even when formal remission is not documented.
What to do with this
- Start with your prescriber, before the first fast. Ask specifically which of your medicines could cause low blood sugar on a low-intake day, and what to change. Don’t stop or skip doses on your own.
- Know how you’d recognize and treat a low. Shakiness, sweating, sudden hunger, confusion and a racing heart can all signal hypoglycemia. Agree on a treatment plan with your care team before you start, and break the fast to treat a low; don’t try to push through it. Our list of who should check with a doctor before fasting covers other situations that need a conversation first.
- Choose a plan with your care team. Ask whether a dietary change is appropriate and what support it would require. After a lifestyle intervention, the consensus advises waiting at least six months to assess remission, alongside the medication-free interval. These trials do not establish that 5:2 causes faster weight loss than 16:8. Our comparison of 16:8, OMAD and 5:2 schedules lays out the trade-offs.
- Judge it by HbA1c, on the right schedule. Ask when to retest. For confirming continued remission, the panel suggests intervals of three months to a year. This does not replace the more frequent glucose monitoring your care team may prescribe.
- Keep your screening appointments. Remission doesn’t cancel eye, kidney and foot checks.
Atomic Fast records fasting start and end times. Those records can help describe your eating pattern during a conversation with your care team. Its phase labels, including “Blood Sugar Normalizing,” are based only on the time since your fast started. They don’t measure your glucose and can’t warn you about a low. If you use a glucose meter or monitor, follow its readings and your care plan, not the timer.
Frequently asked questions
Can intermittent fasting cure type 2 diabetes? No. Some people reach remission after structured fasting-style diets that produce weight loss, but remission can end, so monitoring continues.
Does 16:8 fasting put diabetes into remission? The six-month 16:8 trial discussed here did not measure remission. It found improved HbA1c relative to control, but did not establish equivalence to calorie restriction or show that a daily window produces medication-free remission.
How much weight do I need to lose for remission? There’s no single number. In DiRECT, remission became more common the more weight people lost; in a Chinese trial of leaner adults, it came with an average loss of about 6 kg.
Can I fast if I take metformin? Ask your prescriber before changing your eating pattern. The answer depends on your whole medication list, glucose control and other health needs, not just whether one tablet is metformin. Do not skip a dose to make a fasting schedule work.
Is it safe to fast if I take insulin or a sulfonylurea? A care plan is necessary, but does not guarantee that fasting is suitable for you. These medicines can cause hypoglycemia on low-intake days; trials used clinician-managed dose changes and close monitoring. Never reduce or stop them on your own.
Sources
Sources checked: November 14, 2026.
- Riddle MC et al. (2021), Consensus Report: Definition and Interpretation of Remission in Type 2 Diabetes, Diabetes Care 44(10):2438–2444 — Full text. Supports the remission definition, the three- and six-month timing rules, retesting intervals, the choice of “remission” over “cure”, continued complication screening and the retinopathy caution. Expert consensus, not a trial; the 6.5% cut point itself is flagged for further validation.
- Yang X et al. (2023), Effect of an Intermittent Calorie-restricted Diet on Type 2 Diabetes Remission: A Randomized Controlled Trial, Journal of Clinical Endocrinology & Metabolism 108(6):1415–1424 — Publisher full-text PDF text checked. PMID 36515429; ChiCTR2000038036. 72 adults; remission 17/36 vs 1/36 at the first follow-up, 16/36 at 12 months; about 6 kg mean loss; mean BMI about 24. Small, open-label, supplied kit with medicinal-food ingredients, registration after recruitment began, and nine withdrawals; continuous analyses used 63 participants. No serious adverse events reported, not proof of no hypoglycemia.
- Guo L et al. (2024), A 5:2 Intermittent Fasting Meal Replacement Diet and Glycemic Control for Adults With Diabetes: The EARLY Randomized Clinical Trial, JAMA Network Open 7(6):e2416786 — Full text. PMID 38904963; ChiCTR2000040656. 405 newly diagnosed adults without recent diabetes drugs; HbA1c changes, adjusted differences, weight, week-16 and week-24 proportions below 6.5%, adverse events and funding. Sixteen weeks; not a remission endpoint; funded by the meal-replacement manufacturer. The abstract, results and figure legend report inconsistent confidence intervals; disputed intervals are omitted here.
- Lean MEJ et al. (2018), Primary care-led weight management for remission of type 2 diabetes (DiRECT): an open-label, cluster-randomised trial, The Lancet 391(10120):541–551 — Accepted full manuscript checked in the University of Glasgow repository. PMID 29221645. 306 recruited, 149 per arm analyzed; 46% vs 4% remission at 12 months; remission by weight-loss category. Formula diet, not fasting; clinician-led drug withdrawal began on day one. The 285-person weight categories are observational groupings; remission used the older two-month medication-free definition.
- Pavlou V et al. (2023), Effect of Time-Restricted Eating on Weight Loss in Adults With Type 2 Diabetes: A Randomized Clinical Trial, JAMA Network Open 6(10):e2339337 — Full text. PMID 37889487; NCT05225337. 75 adults, 8-hour window vs calorie restriction vs control for six months; supports the HbA1c and weight comparisons. Small, short, self-reported intake; remission not measured.
- Carter S, Clifton PM, Keogh JB (2018), Effect of Intermittent Compared With Continuous Energy Restricted Diet on Glycemic Control in Patients With Type 2 Diabetes: A Randomized Noninferiority Trial, JAMA Network Open 1(3):e180756 — Full text. PMID 30646030. 137 adults, 12 months; supports the HbA1c equivalence result, the protocol change made because of hypoglycemia, and the 16 of 46 early glucose events. 97 completed; finger-prick monitoring may have missed events.
- NHS, Low blood sugar (hypoglycaemia) — Official clinical guidance supporting symptoms, meal-related risk with insulin and sulfonylureas, prompt treatment and care-team planning. General guidance, not an individualized fasting or dosing plan.
Product note, checked November 14, 2026: the timer’s phase labels are assigned from elapsed time only, as recorded in Atomic Fast’s product documentation and source. This was a source review, not a mobile runtime test, and the app is not a medical device.
Track your eating window and meals with Atomic Fast on the App Store or Google Play.
This article is for general education and isn’t medical advice. If you’re pregnant, managing a health condition, taking medication, have a history of disordered eating, or are considering fasts longer than 24 hours, talk to a healthcare professional first.