Prediabetes Exercise: What the Evidence Actually Supports

Aerobic exercise currently has the best evidence as a single exercise modality for improving insulin resistance and HbA1c in prediabetes. That is a provisional ranking, not a final verdict. The best direct comparison is a network meta-analysis with small resistance-training arms, while a 2025 umbrella review found inconsistent results across exercise modalities and outcomes.

The practical conclusion is less glamorous than a modality cage match: choose an activity you can repeat for at least 12 weeks, make aerobic work the foundation, and add resistance training if you can sustain it. Expect body-composition changes before assuming your blood-sugar markers will follow on the same schedule.

Aerobic training ranked first, not resistance training

In a 2022 network meta-analysis by Li Huang et al., aerobic exercise training ranked first for HOMA-IR and HbA1c among adults with prediabetes. Combined aerobic-plus-resistance training ranked second, resistance training alone ranked third, and control ranked fourth.

The analysis included 13 randomized trials and 567 participants. Across the network, 249 participants were in control groups, 158 received aerobic training, 89 received resistance training, and 71 received combined training. Most trials lasted 12 weeks.

That hierarchy comes from frequentist P-scores. A P-score estimates how likely an intervention is to outperform an average competing intervention across the network. It does not prove that aerobic training beat resistance training by a statistically significant margin in a direct head-to-head comparison. The available report does not provide the aerobic-versus-resistance HOMA-IR estimate and confidence interval needed to make that claim.

The small resistance and combined-training groups matter. With only 89 participants in the resistance-training arm and 71 in the combined arm, estimates comparing those approaches can be imprecise. The network also relied more heavily on control comparisons: the control arm was 2.8 times larger than the resistance-training arm. Rankings can look tidier than the underlying evidence feels.

Three months is a reasonable checkpoint for BMI, not a glycemic deadline

A 2024 systematic review and meta-analysis by Yifei Wang et al. gives a more useful answer to a different question: how long does aerobic exercise take to change body composition in people with prediabetes?

Across 10 randomized trials involving 815 participants, aerobic programs produced a statistically significant reduction in BMI at three months: weighted mean difference −1.75 kg/m² (95% CI, −2.79 to −0.71; P = 0.027). The studies included walking, treadmill exercise, aerobic dance, mixed aerobic programs, and one high-intensity interval protocol. Median intensity was about 65% of maximum heart rate, with programs ranging from 50% to 75% HRmax. The review also reported a statistically significant BMI reduction at six months, although the complete confidence interval is not available here.

Do not treat −1.75 kg/m² as a personal forecast. Heterogeneity in the three-month analysis was high (I² = 79.5%), which means results varied widely among trials. Age, baseline fitness, program design, adherence, and other health conditions can all change the result.

More importantly, the review does not fully report duration-stratified results for fasting glucose, two-hour glucose, HbA1c, or HOMA-IR. A three-month BMI change does not establish that three months reliably produces a clinically meaningful glycemic change. Weight-related and glucose-related outcomes can move at different speeds.

Why 12 weeks is a sensible minimum

A 12-week duration has biological support, although it is not a magic threshold. In a 2022 randomized trial by Kargarfard et al., obese adults with impaired glucose tolerance completed 12 weeks of moderate aerobic exercise. The intervention significantly improved glycemic indices, high-sensitivity C-reactive protein, and ROCK2 activity.

ROCK2 is involved in insulin-signaling pathways, and excessive activity has been linked to insulin resistance. Its reduction is a useful biomarker finding that makes an exercise effect biologically plausible. It does not prove that ROCK2 changes caused the glycemic improvements, nor does this single trial establish the best weekly frequency or total exercise volume.

Still, the study supports a practical planning rule: give a program at least 12 continuous weeks before judging whether it has had time to produce measurable metabolic change. Shorter programs may improve fitness or routine without yet showing much on laboratory markers.

Combined training makes sense, but the evidence cannot crown it

Combined aerobic and resistance training ranked second in the Li Huang network analysis. It is also a sensible practical design: aerobic work can improve cardiorespiratory fitness and glucose uptake during activity, while resistance training can build or preserve muscle mass. More muscle provides more tissue that can take up glucose. That rationale is plausible, but physiology does not substitute for a well-powered head-to-head trial.

A 2024 six-week trial in obese prediabetic soldiers found improvements in lipids, insulin resistance, and body measurements with combined resistance and aerobic training. However, participants also received tirzepatide or placebo in a multi-group design. Tirzepatide prevents clean attribution of the results to exercise alone, and six weeks is shorter than the 12-week duration supported by the aerobic trial.

Other movement options also deserve a place in the conversation. A 2025 meta-analysis of traditional Chinese exercise, including tai chi, qigong, and baduanjin, reported improvements in glycemic markers and psychological outcomes in prediabetes. A 2025 clinical intervention in prediabetic women also reported improved metabolic-health markers with core exercise. These findings do not overturn the aerobic ranking, but they make one point clear: the useful choice is broader than treadmill versus barbell.

The literature still disagrees on how much glucose control exercise delivers

The most important brake on confident exercise prescriptions comes from a 2025 umbrella review by Poon et al.. After synthesizing meta-analyses of randomized exercise trials in prediabetes, the authors concluded that varied modalities and outcomes had produced inconsistent findings that hinder clinical guidance.

Wang et al. also cite Hrubeniuk et al. for a more pointed concern: exercise-related improvements in glucose tolerance in prediabetes were not considered substantial, possibly because interventions were short. That result is secondhand in the available publication, so its sample size, effect estimates, and confidence intervals cannot be independently assessed here. It is still a warning against promising that a few weeks of exercise will normalize glucose markers.

The apparent conflict is not hard to explain. The studies differ in exercise type, intensity, duration, comparator, diagnostic criteria, and outcome. Some measure BMI, some fasting or post-meal glucose, some HbA1c, and some HOMA-IR. These measures answer related questions, not identical ones. A statistically significant shift in a biomarker also does not tell us whether an individual avoided diabetes.

A practical prescription that matches the uncertainty

For someone with prediabetes who is medically cleared to exercise, the evidence supports starting with aerobic training at roughly 50% to 75% of maximum heart rate, near 65% HRmax when that is a useful guide. Aim for at least three sessions each week and continue for at least 12 weeks. Add resistance training about twice weekly if it fits your schedule and recovery.

• Use aerobic activity as the default single modality because it ranked first for HOMA-IR and HbA1c in the available prediabetes network meta-analysis.
• Use a 12-week review point because a randomized trial found metabolic and inflammatory biomarker changes after that duration.
• Judge progress with repeat clinical measures, not enthusiasm alone. BMI, fasting glucose, HbA1c, and HOMA-IR do not necessarily change together.
• Choose walking, cycling, dance, tai chi, qigong, resistance work, or another repeatable format. Adherence is part of the dose.

Exercise improves metabolic health, but current evidence does not show that one modality reliably reverses prediabetes for most people within a few months. Aerobic exercise has the strongest single-modality ranking; combined training is a reasonable practical choice; and the size and durability of glycemic benefit remain unsettled.