Exercise and insulin sensitivity: what the evidence actually says

Insulin sensitivity describes how readily your muscles take up glucose when insulin signals them to. When it falls, your pancreas releases more insulin to clear the same amount of sugar from your blood, and that pattern can precede type 2 diabetes by years.

Exercise is the most reliable way to change it without medication. The research says something more precise than "exercise helps", though: the improvement arrives within a day of training, it is driven mostly by your recent sessions rather than by months of accumulated fitness, and it fades within days of stopping.

What insulin sensitivity is, and how it is measured

Three measurements appear across the studies below, and they are not interchangeable. The hyperinsulinemic-euglycemic clamp is the reference method: insulin is infused at a fixed rate, glucose is infused to hold blood sugar steady, and the amount of glucose needed shows how sensitive the tissues are. An oral glucose tolerance test (OGTT) is simpler, tracking glucose and insulin in the blood for two hours after a 75 g glucose drink. HOMA-IR is calculated from a single fasting blood sample and is the crudest of the three.

A study can improve one of these and leave another unchanged, which explains much of the apparent disagreement in this field.

How quickly does exercise improve insulin sensitivity?

Within about 24 hours. In a 12-week trial in inactive adults with obesity, participants trained four times a week and were asked to keep their body mass constant, so the results reflect the exercise itself rather than weight loss. Peripheral insulin sensitivity measured by clamp was roughly 20% higher the day after the final session than before training, and peak oxygen uptake rose by about 10% (Ryan et al., 2020).

Frequency appears to matter more than any single session. Ten inactive men completed 60 minutes of walking at about 70% of VO₂max in three patterns: three consecutive days, three alternate days, and one session on its own. Only the three consecutive days lowered the insulin response to a glucose load, cutting insulin incremental area under the curve by 34.9% against no exercise. The single session and the alternate-day pattern did not reach statistical significance (Castleberry et al., 2019).

How fast, how often

What the trials measured about timing

24 hTime from a training session to a measurable change in insulin sensitivity

20%Higher clamp-measured insulin sensitivity the day after the last session, with body mass held constant

3 daysConsecutive walking sessions before the insulin response fell; one session alone did not

34.9%Lower insulin incremental area under the curve on that consecutive-day schedule

Two studies, two methods: a hyperinsulinemic-euglycemic clamp in the first, an oral glucose tolerance test in the second. They agree on the direction and on the timescale (Ryan et al., 2020; Castleberry et al., 2019).

How long does the improvement last?

Not long. In the same 12-week trial, trained participants who then avoided exercise for four days saw insulin sensitivity return to its pre-training value, in both the high-intensity and the moderate-intensity groups (Ryan et al., 2020). Twelve weeks of consistent training did not leave behind a reserve of insulin sensitivity that survived four days of rest.

That is a different pattern from cardiorespiratory fitness, which is built slowly and lost slowly. It suggests treating insulin sensitivity as something you maintain with regular sessions rather than something you achieve and keep.

Does intensity change the result?

Less than you might expect. In that trial, 10 one-minute intervals at 90% of maximum heart rate and 45 minutes of continuous moderate cycling produced similar improvements in insulin sensitivity, despite the large difference in time spent training (Ryan et al., 2020).

Across a wider evidence base, intensity gives a modest edge. A meta-analysis of 50 studies found interval training reduced insulin resistance compared with no exercise (standardized mean difference −0.49) and, more narrowly, compared with continuous training (−0.35). HbA1c fell by 0.19 percentage points and body weight by 1.3 kg against controls, and in people at risk of or living with type 2 diabetes, fasting glucose fell by 0.92 mmol/L (Jelleyman et al., 2015).

A 2025 umbrella review pooling 10 meta-analyses, 76 primary studies and 2,954 participants with type 2 diabetes reported HbA1c reductions of 0.39 to 0.83 percentage points against non-exercise controls, alongside cardiorespiratory fitness gains of 3.35 to 6.38 mL/kg/min (Poon et al., 2025). The authors rated the methodological quality of most included reviews as moderate to low, so those figures describe a direction of effect rather than a precise dose.

Where REHIT fits, and where the evidence is thinner

REHIT (Reduced Exertion High-Intensity Interval Training) is the protocol behind CAROL Bike: a short ride of low-intensity cycling built around two 20-second all-out sprints. Its effect on glucose metabolism has been tested directly, with mixed results that are worth stating plainly.

The first trial trained 29 sedentary men and women three times a week for six weeks, in 10-minute sessions. Insulin sensitivity rose by 28% in the men, while maximal aerobic capacity rose by 15% in men and 12% in women. The women's insulin sensitivity did not change significantly (Metcalfe et al., 2012).

A follow-up in 35 sedentary adults tested whether that was a real sex difference. It was not: responses varied widely between individuals regardless of sex. In that group, aerobic capacity again improved, but insulin area under the curve showed only a trend (p = 0.096) and the Cederholm index of insulin sensitivity did not change (Metcalfe et al., 2016).

In 16 men already living with type 2 diabetes, eight weeks of REHIT was compared with eight weeks of moderate walking that took five times as long. REHIT produced the larger fitness gain, 7% against 1%, and both lowered mean arterial pressure by about 4%. Neither improved OGTT-derived insulin sensitivity or continuous glucose monitoring measures over that period (Ruffino et al., 2017).

Three REHIT trials

Aerobic capacity is the return that repeats

Sedentary menMetcalfe 2012, 6 weeks

+15%

Sedentary womensame trial

+12%

Men with type 2 diabetesRuffino 2017, 8 weeks

+7%

Moderate walkingsame trial, five times the time

+1%

Change in maximal aerobic capacity, bars scaled to the largest gain. Insulin sensitivity moved in only one of these groups, the men in 2012, by 28% — which is the distance between what REHIT reliably does and what it is often assumed to do.

The honest reading: REHIT's most consistent, best-replicated return is cardiorespiratory fitness. Its effect on insulin sensitivity is real in some sedentary groups, absent in others, and not established in people who already have type 2 diabetes. If your goal is glycemic control specifically, that uncertainty is worth knowing before you choose a protocol, and worth discussing with your clinician if you take glucose-lowering medication.

What this means for how you train

Three practical points follow from the evidence rather than from habit.

• Regularity beats any single session. Consecutive days changed the insulin response where one session did not, and four days off erased twelve weeks of gains. Leaving more than a couple of days between sessions gives up most of the benefit.
• You do not need long sessions to move the marker. Ten one-minute intervals matched 45 minutes of continuous cycling in a controlled comparison, and the interval meta-analysis found a small advantage for intensity.
• Expect variation between people. The REHIT trials found large differences in individual response for every outcome measured. A marker that does not move for you does not mean the training failed, particularly when aerobic capacity is improving at the same time.

VO₂max remains the outcome with the strongest and most consistent evidence behind short, hard cycling, and it is also the marker most closely tied to long-term health. Insulin sensitivity is best treated as a second reason to keep the sessions frequent.

The bottom line

Exercise improves insulin sensitivity quickly, by roughly 20% the day after training in a controlled clamp study, and the effect depends on how recently you trained rather than on how long you have been training. Four days without exercise returned it to baseline. Intensity offers a modest advantage over continuous moderate work, and short interval sessions match much longer ones. For REHIT specifically, the fitness evidence is strong and the glycemic evidence is mixed. Train often, and judge the result by more than one number.

This article describes research findings and is not medical advice. If you are managing type 2 diabetes or taking medication that lowers blood glucose, speak to your clinician before changing how you train.