Exercise as an antidepressant: what dose seems to work best?

What the science says

Exercise is not a cure-all for depression, but the clinical literature supports it as a meaningful antidepressant intervention, especially for people with mild to moderate symptoms and for those who can stick with a structured routine[1][11][14]. A recent umbrella-style synthesis in adults and pragmatic clinical studies continue to show symptom improvement, with some programs producing changes that are comparable to standard treatments in selected populations[1][3][13].

The newer and more interesting question is not whether exercise helps, but how much exercise is enough. Dose-response analyses suggest the benefit is not linear in a simplistic “more is always better” way; instead, symptom reduction appears to depend on frequency, intensity, and total program duration[2][3]. In youth, a 2025 meta-analysis found that aerobic exercise significantly reduced depressive symptoms and reported statistically significant dose-response relationships across age, intensity, and duration[2]. In adults, a pragmatic trial of supervised running prescribed three times weekly for 16 weeks found antidepressive benefits alongside metabolic and cardiovascular gains[3].

How it works

The antidepressant effects of exercise likely arise from several overlapping pathways rather than one single mechanism. Reviews of the exercise-depression literature consistently point to changes in neurotransmission, neuroplasticity, inflammation, and stress-hormone regulation[11][12][14]. Exercise may increase brain-derived neurotrophic support, improve synaptic plasticity, and reduce inflammatory signaling that can interfere with mood regulation[11][12].

There is also a behavioral mechanism: exercise can restore structure, routine, self-efficacy, and social contact, all of which matter in depression even when the biological effects are modest[1][14]. That is one reason pragmatic, supervised programs often perform better than loosely advised “be more active” recommendations: the intervention is more observable, more accountable, and easier to sustain[1][3].

The dose-response signal probably reflects both biology and adherence. Too little exercise may not reach a threshold that changes mood-related pathways; too much, especially if it is unrealistic or poorly tolerated, may reduce adherence and blunt overall benefit[2][3]. A clinically useful dose is therefore the one a person can repeat consistently.

What the evidence supports

The evidence base is strongest for aerobic exercise, especially programs that are moderate in intensity, repeated several times per week, and continued for at least a few weeks[2][3][13]. In youth, the 2025 meta-analysis reported that a moderate-intensity aerobic program lasting 25 to 40 minutes, performed three times weekly for 9 to 15 weeks, was associated with greater reductions in depressive symptoms[2]. That is one of the clearest published dose-response signals currently available.

In adults, the best-supported practical model is a supervised aerobic routine that is frequent enough to build habit and sufficiently challenging to reach moderate-to-vigorous effort[3][13][15]. The BMJ Evidence-Based Nursing summary of a pragmatic outpatient study describes a prescription of three supervised outdoor sessions per week, each including warm-up, 30 minutes of running at 70%–85% heart rate reserve, and cool-down over 16 weeks[3]. That kind of structured program is notable because it tests exercise under real-world conditions, not just in idealized laboratory settings.

The totality of the evidence also suggests that exercise can be useful as an adjunct to standard depression care and, in some patients, may help maintain gains after acute treatment[13][15]. However, the literature is still uneven: trials vary widely in exercise type, intensity, supervision, comparator group, and depression severity, which makes exact dose thresholds hard to generalize[1][11][14].

Practical takeaways

• Start with aerobic exercise if the goal is antidepressant benefit, because that is where the dose-response data are clearest[2][3].
• A practical starting target is 25 to 40 minutes, 3 times per week, at a moderate intensity, especially when building a routine from scratch[2].
• If someone is already active, the evidence suggests that structured progression and consistency matter more than chasing maximal intensity[2][3].
• Supervision, scheduling, and social accountability can improve adherence and may help explain why formal exercise prescriptions often outperform vague advice[1][3].
• Exercise should be framed as part of a depression plan, not a replacement for assessment, psychotherapy, medication, or crisis care when those are indicated[1][13][14].

Caveats and unknowns

The biggest limitation in this literature is heterogeneity: studies differ in participant age, baseline fitness, depression severity, intervention format, and how depression is measured[1][2][14]. That makes the dose-response pattern real but not yet precise enough to serve as a universal prescription.

Another limitation is publication and adherence bias. People who can enroll in and complete exercise trials may be more motivated, less medically complex, or better resourced than many patients seen in routine care[1][3][11]. That matters because the “best” dose in a trial may not be the best dose for someone with fatigue, pain, anxiety, caregiving burden, or low access to safe exercise spaces.

It is also still unclear whether certain modalities beyond aerobic exercise—such as resistance training, yoga, or mixed programs—have distinct antidepressant dose-response curves, or whether they work mainly through similar upstream pathways[12][14]. Mechanistic work is promising, but much of it remains preclinical or translational rather than definitive for clinical prescription[10][11][12].

The most defensible conclusion today is that exercise has a real antidepressant signal, and that signal appears to strengthen with a moderate, repeatable, supervised dose rather than with sporadic effort or extreme training[2][3][13].