The use of transcranial magnetic stimulation (TMS) as a therapeutic intervention is FDA-cleared for treating depression, obsessive-compulsive disorder, and migraine, and shows promise for a host of other brain disorders. The appeal of TMS is its safety, non-invasiveness, and well-established capacity for modulating the activity of brain regions. In human subjects, that modulation is assessed only at the gross scale of behavioral, cognitive, or aggregate physiological effects (e.g. EMG, EEG, fMRI). The fine-scale responses and mechanisms of TMS, at the level of biophysical and biological effects on neurons and circuits, remain poorly understood. This knowledge gap hinders rational design of TMS protocols and leaves researchers and clinicians dependent on trial-and-error approaches and inferences from macroscopic data to improve the methodology. The lack of reductionistic insight is particularly detrimental when targeting non-motor areas such as prefrontal cortex where a readout of the immediate neural response is unavailable, for example due to the stimulus artifact in EEG. Our overall goal is to fill in this knowledge gap by studying the neural circuit mechanisms of TMS in the non-human primate brain. The approach integrates neurophysiological experiments featuring direct single-unit and local field potential recordings and multiscale computational simulations of neural circuits in both primary motor cortex and prefrontal cortex. Aim 1 is to establish the circuit mechanisms of acute responses to single and paired TMS pulses. Determining the pulse response of single neural elements and recurrent cortical circuits permits a detailed examination of the biophysics and biology of neural recruitment at a short time scale. A main goal of TMS therapy is to achieve controlled, lasting neuromodulation, however, so in Aim 2 we will extend the same neurophysiological and modeling approaches to the study of responses to repetitive TMS (rTMS). Here the goal of the neurophysiology will be to quantify the effects of rTMS pulse trains on long-lasting changes in neural activity and, accordingly, the neural simulations will incorporate synaptic plasticity. Critically, in both Aims we will conduct the experiments and modeling both in primary motor cortex, where spinal potential recordings and electromyography can supplement direct readout of neural effects in cortex, and prefrontal cortex, where only cortical-level recordings are suited to characterize neuromodulatory effects. The overall product will be an experiment- and model-driven mechanistic understanding of the effect of TMS on cortical circuits, enabling a transformational advance in the interpretation of the effects of TMS. Taken together, the results will promote a more biologically-grounded, rational approach to designing TMS protocols for neuromodulation.