"Exercise more" is not a plan. This analysis turns it into actual minutes. Researchers pooled 39 randomized controlled trials, 3421 participants with prediabetes (average age 55.8, average BMI 26.85, studies from 12 countries), to work out not just whether exercise helps blood sugar control, but how much of which kind is actually needed.
Every exercise type helped, but not equally. Compared with usual care, high-intensity interval training (HIIT) produced the largest average drop in HbA1c (-0.78 percentage points), followed by combined training — cardio plus strength — (-0.36), resistance training alone (-0.30), aerobic training alone (-0.28), and mind-body practices like yoga or tai chi (-0.19). But there's a catch with the HIIT number: it's rated very low certainty evidence, and HIIT also had the lowest adherence of any group in the trials. Combined training, resistance training, and aerobic training all carried moderate-certainty evidence and combined training had the best adherence of the five.
The overall sweet spot was 850 MET-minutes per week — a nonlinear, J-shaped relationship, meaning benefits rise with dose up to a point rather than in a straight line. Translated into actual minutes at different intensities, for someone at a normal or overweight BMI:
- Light activity (like a leisurely walk): roughly 293–531 minutes per week for the optimal effect
- Moderate activity (brisk walking, cycling): roughly 144–283 minutes per week
- Vigorous activity (running, HIIT-style effort): roughly 142 minutes per week
A brief workout here and there, well short of these ranges, is unlikely to move the needle on blood sugar — the harder the effort, the less total time is needed, but there's still a floor to clear.
Intensity mattered more than duration alone. At a fixed 150 minutes per week — the standard public health guideline — HbA1c improvement only became statistically significant once intensity reached about 2.8 METs, and only crossed the threshold considered clinically meaningful at around 8.3 METs (solidly vigorous effort). Broken into standard categories: light activity (1.6–2.9 METs) showed no significant effect at all, moderate activity (3.0–5.9 METs) gave a modest reduction (-0.15%), and vigorous activity (≥6.0 METs) gave the biggest reduction (-0.28%).
Body weight changed the picture too. At the optimal dose, only people with obesity saw a reduction large enough to be considered clinically meaningful (-0.339%). People at normal weight or overweight saw statistically real but smaller improvements that didn't cross that clinical threshold, even at the optimal dose. For people with obesity specifically, meaningful benefit came at a lower dose too — around 550 MET-minutes per week, rather than 850.
What this can and can't tell you. Risk of bias was low in only 17.9% of the 39 trials (unclear in 69.2%, high in 12.8%), and the HIIT and mind-body findings in particular carry very low certainty. There was no evidence of publication bias skewing the results (Egger's test p = 0.77). No conflicts of interest were declared.
Bottom line. For prediabetes, the exercise dose that mattered most in this analysis was around 850 MET-minutes per week, achievable as roughly 142 minutes of vigorous activity, 144–283 minutes of moderate activity, or 293–531 minutes of light activity per week — with combined training (cardio plus strength) offering the best balance of solid effect and real-world adherence.
Medical note. This article is for general education purposes only and should not be considered personal medical advice.