Exercise for Fatty Liver: Best Type and Minimum Dose

Science Finally Put a Number on Exercise for Fatty Liver

For years, the advice for NAFLD, now usually called MASLD, has been vague: exercise more. That is better than nothing, but it is not a prescription. Two network meta-analyses published in 2025 and 2026 finally moved the field closer to one. The most useful number comes from Tan et al. 2026, a Bayesian network meta-analysis with a dose-response model in 24 randomized controlled trials involving 961 adults with MASLD. Their model estimated that combined aerobic-resistance training reached the minimum clinically significant threshold for reducing intrahepatic lipid at 130 MET-minutes per week. Using the paper's own conversion of 4.5 METs for moderate-intensity activity, that works out to about 29 minutes per week total.

That is a model, not a stopwatch-tested trial result. No individual RCT in the analysis assigned people to exactly 29 minutes a week and then measured the liver response. Still, the finding matters because it replaces hand-waving with a quantitative threshold. The same analysis estimated a broader minimum effective dose for exercise overall at about 460 MET-minutes per week, a cost-effective target at about 630 MET-minutes per week, and a plateau around 850 MET-minutes per week. At 630 MET-minutes per week, the standardized mean difference for hepatic steatosis was -0.68 with a 95% confidence interval from -0.95 to -0.40. At 850 MET-minutes per week, the SMD was -0.72 with a 95% confidence interval from -0.97 to -0.49. The extra gain above 630 was small.

If you want the practical point: the smart target is not "as much as possible." It is roughly the cost-effective zone, because the plateau buys little.

Why One Meta-Analysis Picks HIIT and Another Picks Combined Training

The second key paper, Khalafi et al. 2025, looked at 38 randomized trials with 1,880 adults with and without NAFLD and ranked high-intensity interval training first for reducing intrahepatic lipid content, with a p-score of 0.95. Combined aerobic-resistance training ranked third, with a p-score of 0.52. The overall exercise effect was an SMD of -0.33, reported as statistically significant with p=0.001.

That clashes with Tan et al., who ranked combined aerobic-resistance training first with a SUCRA value of 0.81. The cleanest explanation is methodological, not biological drama. Khalafi et al. pooled comparative effects across studies but did not standardize for exercise volume actually delivered. Tan et al. added a non-linear dose-response layer and asked a different question: which modality reaches meaningful liver fat reduction most efficiently per unit of exercise dose?

Under that framework, combined training came out ahead. Its modeled optimal dose was 640 MET-minutes per week. Moderate-intensity continuous training came in at 590 MET-minutes per week, high-intensity circuit training at 880, and resistance training needed more than 640 MET-minutes per week just to enter the clinically meaningful range. HIIT had a different profile in Tan et al.: benefits kept rising without a defined plateau.

That does not prove HIIT is inferior. It means HIIT may look best when trials deliver larger or more tightly supervised workloads, while combined training may look best when the analysis cares about dose efficiency. Those are not the same question, and patients live in the gap between them.

What the Minimum Dose Probably Means in Real Life

The 130 MET-minute threshold for combined training is the headline number because it is startlingly low. It is also the number most likely to be misunderstood. Tan et al. derived it from a Bayesian posterior distribution across heterogeneous trials, not from a single trial that directly observed the threshold. Low-dose estimates in these models usually carry wide uncertainty, and the paper's value should be read as a plausible minimum threshold, not a guaranteed floor that works for everyone.

The sturdier number for actual prescribing is the cost-effective range around 630 to 640 MET-minutes per week. Tan et al. translate that to something like 30 minutes of moderate-intensity activity five days per week, or 30 minutes of vigorous activity three days per week. For a sedentary patient with MASLD, that looks less like an athlete's program and more like a realistic weekly structure: some aerobic work, some resistance work, enough total volume to hit the target, and no need to chase the plateau.

The broad ranking, based on Tan et al.'s dose model, looks like this: combined aerobic-resistance training first for dose efficiency, MICT next, HICT requiring more total dose, HIIT potentially effective across increasing doses but without a defined plateau, and resistance training alone as the least dose-efficient option for liver fat.

Exercise Still Works Even If the Scale Refuses to Cooperate

A separate question matters just as much for patients: does exercise help the liver only because it causes weight loss? Babu et al. 2021 tackled that directly in a systematic review and meta-analysis restricted to exercise interventions without dietary modification or significant weight reduction. Their conclusion was straightforward: exercise improved NAFLD-related clinical parameters even without weight loss.

That changes the counseling. If a patient exercises and the body weight barely moves, the intervention can still be working where it counts for this disease. The liver is responding to the exercise signal itself, not just to fewer calories. That does not make weight loss irrelevant. It does mean the absence of weight loss is not proof of failure.

Mechanistically, Tan et al. and Khalafi et al. do not settle why combined training may be more dose-efficient. One reasonable hypothesis is that aerobic exercise and resistance exercise hit partly different pathways at the same time: aerobic work may favor hepatic fat oxidation and mitochondrial adaptations, while resistance work may alter myokine signaling from skeletal muscle. But that remains an inference. No mechanistic RCTs in the cited evidence directly compared these pathways across exercise modalities in MASLD. The ranking should rest on the meta-analyses, not on a story about AMPK and irisin that sounds good on a slide deck.

Exercise Alone Helps. Exercise Plus Diet Helps More.

None of this means exercise is the whole treatment. Artola Arita et al. 2025 compared Mediterranean diet alone, exercise alone, and their combination in adults with MASLD and found that the combination outperformed either one alone for body composition and liver outcomes. Bagnato et al. 2024 adds randomized clinical trial evidence that healthy lifestyle changes improved liver steatosis in patients with MASLD.

So exercise alone is not the ceiling. It is the floor. That matters because some readers will hear "29 minutes a week" and imagine a loophole big enough to drive a blender bottle through. The evidence does not support that. It supports a minimum threshold for a liver fat signal, a cost-effective target for a more meaningful effect, and better outcomes still when diet joins the program.

The Limits of the Evidence

The biggest limitation is simple: no head-to-head randomized trial in the cited evidence directly compared HIIT with combined aerobic-resistance training using liver fat as the primary endpoint. Both rankings come from indirect network comparisons through shared control arms. That is useful, but it is not the same as direct evidence.

There are other problems. Tan et al.'s dose thresholds are modeled rather than directly observed at those exact exercise volumes. Khalafi et al. included adults with and without NAFLD, and the paper did not separately report NAFLD-only subgroup results for the modality ranking. Neither primary network meta-analysis reported modality-specific adherence percentages or dropout rates in a way that lets readers compare real-world effectiveness. That matters a lot for HIIT, because a protocol that wins in a supervised trial can lose badly once regular life starts interrupting it.

Chin et al. 2022 is relevant here, but carefully so. It is a protocol paper, not an outcomes paper. The authors proposed a one-year, once-weekly HIIT trial in adults with central obesity and included liver fat as a primary outcome. The design is interesting because it was explicitly framed as a more feasible format for sedentary populations. What it does not provide is completed liver fat data. It shows that the adherence problem is recognized. It does not solve it.

One more caveat: both primary network meta-analyses are vulnerable to the weight-loss confound. If exercise caused some participants to lose weight, then liver fat reduction may reflect both exercise-specific effects and negative energy balance. Babu et al. partially addresses that issue by focusing on no-weight-loss trials, but that narrower approach may also select for lighter protocols and smaller samples.

What a Sensible Prescription Looks Like Now

The evidence no longer supports giving patients a generic order to move more and hoping for the best. It supports a more specific hierarchy.

• Exercise reduces liver fat in NAFLD/MASLD, and the effect is clinically meaningful at adequate doses.
• Exercise can improve liver-related outcomes even without weight loss.
• Combined aerobic-resistance training looks most dose-efficient in the 2026 Bayesian model.
• A practical target is about 630 to 640 MET-minutes per week, because gains beyond that appear to flatten.
• HIIT remains a valid option and ranked first in the 2025 dose-agnostic network meta-analysis, but that edge may reflect trial design and delivered volume more than superiority per unit dose.

The best current summary is boring in the way useful medicine is often boring. Do some exercise. If possible, do enough to reach roughly 630 MET-minutes per week. If you want the most efficient format from the best dose-response model we have, combine aerobic and resistance work. Add a Mediterranean-style diet if you can. And if someone promises a magic modality with no caveats, check whether they have an actual head-to-head trial or just a very confident voice.

Current evidence supports exercise as treatment for MASLD, but the "best" modality still rests on indirect comparisons. The dose target is getting clearer faster than the modality winner.