Reassessing the neurophysiological effects of repetitive transcranial magnetic stimulation: A systematic review and comparative meta-analysis across protocols, outcome measures and cortical sites

Repetitive transcranial magnetic stimulation (rTMS) is widely used to modulate brain function and behavior. Initial studies on its neurophysiological effects established a binary model of cortical excitability modulation derived primarily from studies on motor cortex, with stimulation frequency determining excitatory or inhibitory outcomes. However, whether these assumptions hold across protocols, outcome measures, and cortical targets remains unclear. We provide a comprehensive meta-analysis spanning 2014-2025, comparing effects across domains and examining key moderators. Our systematic search identified 202 studies involving 473 datasets and 8079 healthy individuals. Qualitative and quantitative analyses were conducted on most commonly used protocols traditionally presumed to be "excitatory" (high-frequency rTMS (≥5Hz), intermittent theta-burst stimulation (iTBS)) and "inhibitory" (low-frequency rTMS (<5Hz), continuous TBS (cTBS)). rTMS induced changes in corticospinal, paired-pulse and cortical excitability (TMS-EEG) measures were investigated, supplemented by subgroup and meta-regression analyses. Motor evoked potential (MEP) amplitude was moderately facilitated by "excitatory", and weakly suppressed by "inhibitory" protocols, with no significant difference between conventional high/low-frequency and TBS paradigms. Effects of "inhibitory" protocols did not survive sham-normalization. Meta-regression revealed decreasing effects over years, driven by greater methodological rigor and neuronavigation use. Critically, no consistent effects were found for paired-pulse or TMS-EEG metrics for any protocol. Single-session rTMS effects are weaker and more variable than previously assumed, and the binary model derived from MEPs does not generalize to other excitability measures or non-motor regions. Future work should avoid heuristics, and prioritize mechanistic validation using direct cortical measures, sham-controlled and test-retest designs, and brain-state guided individualization.