Insulin falls because you stopped eating — that part is arithmetic#
Fasting lowers your insulin. This is the least contested claim in the whole field and also the least interesting one, because insulin rises in response to food and you have removed the food. Twenty-two days of alternate-day fasting cut fasting insulin by 57%; that is not a discovery about fasting, it is a description of not eating. The claim worth examining is the one stacked on top: that lowering insulin exposure also improves insulin sensitivity — how well your muscles respond to the insulin you do produce.
On that second question the evidence runs in both directions at once, and not because the studies are sloppy. Over hours to a day or two, fasting actively induces insulin resistance in muscle, deliberately, as a mechanism for keeping glucose available to the brain. Over weeks to months, fasting protocols that produce weight loss improve insulin sensitivity roughly as much as the weight loss predicts. Both are real physiology on different clocks, and confusing them is why the internet argument never resolves. Here is what actually happens, hour by hour and then month by month.
What your blood sugar does across a fast#
The reassuring part first: your glucose does not fall off a cliff, and the system that prevents it is well characterized. In the hours after your last meal, insulin falls toward basal and the liver releases stored glycogen to hold blood glucose steady. When glycogen runs low, the body switches to making glucose from scratch — gluconeogenesis, drawing chiefly on amino acids from muscle protein — while most tissues move onto fat-derived fuels, free fatty acids and the ketone bodies made from them1.
That is the architecture of a fast. Glucose is defended, not abandoned, because a handful of tissues still need it; everything that can burn fat is pushed onto fat so that the glucose being defended goes further. The timing of that fuel switch — and why a 16-hour window barely reaches its near edge — belongs to the intermittent fasting pillar.
What matters here is the consequence nobody advertises. Sparing glucose for the brain requires stopping muscle from taking it up. Muscle is the largest consumer of glucose in your body, and the way the system shuts that consumption down is by making muscle stop responding properly to insulin.
The counterintuitive part: a long fast makes you temporarily insulin resistant#
This has been measured directly, and the effect is not subtle.
Fourteen healthy young adults were studied on three occasions in random order: after an overnight fast, after a 24-hour fast with placebo, and after a 24-hour fast with acipimox, a drug that blocks the release of fatty acids from fat tissue. The 24-hour fast raised overnight free fatty acids 2.8-fold. Insulin sensitivity fell from 5.7 to 2.6 (×10⁻⁴ min⁻¹ per mU/l, P < 0.0001) — less than half its starting value. The acute insulin response also fell (P = 0.02), so the disposition index, which combines the two, dropped from 93.1 to 35.5 (P < 0.0001)2.
Go longer and it deepens. In six healthy lean men fasted for 60 hours, insulin-mediated peripheral glucose uptake fell by 46% (P = 0.03)3.
The Salgin design contains the mechanism as well as the finding, which is why it is the better of the two studies. When the researchers gave acipimox during the fast, fatty acids fell by an average of 20% and the disposition index rose 31% (P = 0.03) — a partial reversal. So this is not damage that fasting inflicts on the insulin system. It is substrate competition: with fat pouring into circulation, muscle burns fat and declines glucose, and the insulin resistance is the readout of that choice. Lower the fatty acids and it partly goes away.
Hold the scale of these studies in view — fourteen and six young, healthy, lean people, in acute laboratory conditions. They establish direction and mechanism, not magnitude in anyone else. But the direction is unambiguous, and it is the opposite of what the popular claim implies.
So how does fasting also improve insulin sensitivity?#
The two findings are not in conflict once you separate three things the word "fasting" runs together.
Timescale. The resistance above is measured during the fast. It resolves when you eat. The improvements people report are measured weeks or months in, in the fed state, after body composition has changed.
Cause. Most of the long-run improvement travels with fat loss, which improves insulin sensitivity whether you achieved it by fasting or by any other means. The narrow, genuinely interesting question — whether an eating window improves these markers with weight held constant — has been tested in a small supervised-feeding trial and a larger null one, and both are laid out in time-restricted eating benefits.
Tissue. The fasting-induced resistance is peripheral, in muscle. Hepatic insulin action moves differently. "Insulin sensitivity" is not one number.
| The claim | What the evidence actually supports |
|---|---|
| Fasting lowers insulin | Yes — mechanically, because you removed the food |
| Fasting improves insulin sensitivity during the fast | No — a 24-h fast more than halved it2 |
| A long fast reduces muscle glucose uptake | Yes — down 46% at 60 hours3 |
| Fasting improves long-run glycemic control | Not beyond what the diet achieves4 |
That last row is the clinical bottom line, and it is worth stating plainly rather than hedging. Pooling 11 randomized trials in people with type 2 diabetes, intermittent fasting produced no significant difference in HbA1c against control (SMD −0.08, 95% CI −0.20 to 0.04; p = 0.19), and none in fasting blood glucose across seven studies (SMD 0.06, 95% CI −0.25 to 0.38; p = 0.69). The reviewers concluded that intermittent fasting and a usual diet pattern "have no difference in terms of glycaemic control"4. This is a small review in a specialist journal rather than a Cochrane-grade verdict, and the confidence intervals are wide enough to hide a modest real effect — but nothing in it supports fasting as a glucose treatment distinct from the weight loss it produces.
Why your fasting glucose can read higher, not lower#
This is the question people actually arrive with, usually after watching a continuous glucose monitor through a long fast and finding the number stubbornly up rather than down.
The physiology above explains it without anything having gone wrong. Once glycogen is depleted, the liver is manufacturing glucose continuously while muscle — the tissue that would normally clear it — has stopped taking much up. Production stays on and disposal is throttled, so the circulating level can sit at or above where it started, with insulin low rather than high. A glucose reading that looks like insulin resistance is, in that moment, exactly what insulin resistance looks like, and it is doing its job.
Two practical consequences follow. A glucose tolerance test taken after an unusually long fast is not measuring your ordinary metabolism, and it will read worse. And short-term readings during a fasting protocol are a poor guide to whether the protocol is working — the marker that tracks the thing you care about is what happens to your fasted numbers after weeks of fat loss, not what your meter says at hour 30. Poor sleep, meanwhile, moves the same markers harder over a single night than a shifted meal does, as sleep and insulin sensitivity sets out, and the food you break the fast with matters more than the fast's length — a fast-emptying carbohydrate load lands on a system primed to handle it badly, which is where the caveats around glycemic index actually earn their keep.
If you take glucose-lowering medication#
Everything above describes a system with intact counter-regulation. Insulin and sulfonylureas lower glucose on a schedule set by the dose rather than by whether food arrived, so the same physiology that is uneventful in a healthy person becomes a hypoglycemia risk — one that persists even when doses have been reduced in advance. The trial evidence with an actual rate ratio on it, plus the other groups for whom the answer changes, is in who should not fast. The sequence is a medication review with your clinician first and a fasting schedule second, never the reverse.
FAQ#
Why did my fasting blood sugar go up during a long fast?#
Because your liver keeps producing glucose while your muscles temporarily stop taking it up — a deliberate mechanism for keeping glucose available to the brain. After 60 hours without food, insulin-mediated glucose uptake in the periphery fell 46% in healthy men3. Production continues, disposal is throttled, and the reading holds or rises with insulin low. It resolves when you eat.
Does fasting lower insulin?#
Yes, reliably and trivially — insulin responds to food, and you have removed it. Whether that translates into better insulin sensitivity is the separate question, and during the fast itself it moves the other way: a 24-hour fast more than halved insulin sensitivity in healthy adults, driven by a 2.8-fold rise in circulating fatty acids2. Long-run improvements largely track the fat you lose.
Does intermittent fasting improve HbA1c in type 2 diabetes?#
Not beyond what any equivalent diet achieves, on the pooled evidence. Across 11 randomized trials, intermittent fasting produced no significant HbA1c difference against control (SMD −0.08, 95% CI −0.20 to 0.04) and no difference in fasting glucose4. It remains a legitimate way to run a deficit — it is just not a glucose intervention in its own right, and it interacts with medication in ways that need supervising.
Sources#
- Cahill GF Jr. Fuel metabolism in starvation. Annu Rev Nutr. 2006;26:1-22.
- Salgin B, Marcovecchio ML, Humphreys SM, et al. Effects of prolonged fasting and sustained lipolysis on insulin secretion and insulin sensitivity in normal subjects. Am J Physiol Endocrinol Metab. 2009;296(3):E454-E461.
- van der Crabben SN, Allick G, Ackermans MT, Endert E, Romijn JA, Sauerwein HP. Prolonged fasting induces peripheral insulin resistance, which is not ameliorated by high-dose salicylate. J Clin Endocrinol Metab. 2008;93(2):638-641.
- Sharma SK, Mudgal SK, Kalra S, Gaur R, Thakur K, Agarwal R. Effect of Intermittent Fasting on Glycaemic Control in Patients With Type 2 Diabetes Mellitus: A Systematic Review and Meta-analysis of Randomized Controlled Trials. touchREV Endocrinol. 2023;19(1):25-32.
