Exercise and BDNF: How Training Supports Brain Plasticity

What Is BDNF and Why It Matters for the Brain

BDNF (brain-derived neurotrophic factor) is a protein from the neurotrophin family that supports neuronal survival, enables synaptic plasticity, and stimulates neurogenesis in the hippocampus. The hippocampus is the structure responsible for forming episodic and spatial memory.

BDNF acts as a molecular intermediary between physical activity and brain function. During exercise, working muscles release irisin and other myokines that cross the blood-brain barrier and stimulate BDNF synthesis in the hippocampus. At the same time, cerebral blood flow increases, improving delivery of oxygen and nutrients to neurons.

Reduced plasma BDNF levels are observed in Alzheimer's disease, depression, and schizophrenia. This makes BDNF one of the most studied biomarkers of neuroplasticity, although the causal relationship between its levels and clinical outcomes remains an active area of research.

Exercise and the Hippocampus: RCT Evidence

The landmark evidence that exercise changes brain structure comes from Erickson et al., published in PNAS in 2011. The researchers randomised 120 older adults (mean age 67, without dementia) to one year of aerobic exercise (walking, 3 times per week, progressed to moderate intensity) or to stretching and toning exercises.

Results at 12 months:

• Left hippocampal volume increased by 2.12% in the exercise group — and decreased by 1.4% in the control group.
• Right hippocampal volume increased by 1.97% in the exercise group — versus a decrease of 1.4% in controls.
• Serum BDNF levels rose significantly in the exercise group.
• Spatial memory (dot memory task) improved in the exercise group.

The relationship between hippocampal volume gains and BDNF increases was statistically significant. The authors interpreted this as confirming BDNF's role in mediating structural brain changes with aerobic exercise. This RCT with MRI outcomes in older adults is among the most cited in exercise neuroscience.

BDNF Meta-Analyses: Which Exercise Type Is Most Effective?

The meta-analysis by Karimian et al. (Brain Sciences, 2025), which included 19 randomised controlled trials with a total of 850 participants (443 in intervention groups, 407 in controls), evaluated the effect of regular exercise on peripheral BDNF in neurological and general populations.

Overall effect: SMD = 1.03 (95% CI 0.56–1.49, p < 0.0001) — large by Cohen's criteria. Exercise type ranking by effect size:

• Resistance training: SMD = 1.57 (95% CI 0.91–2.23)
• Aerobic exercise: SMD = 1.44 (95% CI 0.36–2.52)
• Combined protocols: SMD = 0.55 (95% CI 0.21–0.89)

By duration: the greatest BDNF increase was seen with 8–12 week interventions (SMD = 1.42, 95% CI 0.84–2.00). Programmes longer than 12 weeks produced a smaller effect (SMD = 0.68), which the authors attribute to receptor adaptation. The authors rated evidence quality using the GRADE system as low — due to high sample heterogeneity and differences in BDNF measurement protocols.

Aerobic Exercise in Older Adults: Which Intensity Works Best?

A separate network meta-analysis by Zhu et al. (Frontiers in Aging Neuroscience, 2025), including 17 RCTs with 900 older adults (mean age 69), compared three aerobic exercise modalities: walking, running, and cycling.

Overall effect on BDNF levels: SMD = 0.62 (95% CI 0.06–1.18, p = 0.03) — moderate. The most effective protocol by network analysis (SUCRA 99.9%) was low-intensity short-duration walking. Moderate-intensity walking ranked second (SUCRA 83.7%). High-intensity long-duration protocols were less effective for BDNF in this age group.

A moderate baseline BDNF level predicts a stronger response to exercise (beta = −0.0395, p = 0.026) — meaning people with lower baseline levels have greater potential for increase. Older age positively predicted the response to intervention (beta = 0.0298, p = 0.026).

Caveats: What the Data Do Not Allow Us to Claim

Peripheral BDNF (plasma or serum) is an indirect measure of central BDNF in the brain. Both correlate but are not identical. Most clinical meta-analyses use peripheral BDNF as a surrogate marker. Translating "BDNF rose by X pg/mL" into "memory improved" requires separate RCTs with cognitive outcomes — these are fewer in number and more heterogeneous.

Furthermore, some BDNF changes described in meta-analyses occur in clinical samples (neurodegenerative diseases, depression, type 2 diabetes). Applying the SMD magnitude to healthy middle-aged adults is an assumption that the authors themselves describe as a source of heterogeneity.