To better understand the neuronal effects of transcranial magnetic stimulation (TMS), we studied how the TMS-evoked brain responses depend on stimulation intensity. We measured electroencephalographic (EEG) responses to motor-cortex TMS, estimated the intensity dependence of the overall brain response, and compared it to a theoretical model for the intensity dependence of the TMS-evoked neuronal activity. Left and right motor cortices of seven volunteers were stimulated at intensities of 60, 80, 100, and 120% of the motor threshold (MT). A figure-of-eight coil (diameter of each loop 4 cm) was used for focal stimulation. EEG was recorded with 60 scalp electrodes. The intensity of 60% of MT was sufficient to produce a distinct global mean field amplitude (GMFA) waveform in all subjects. The GMFA, reflecting the overall brain response, was composed of four peaks, appearing at 15 ± 5 msec (Peak I), 44 ± 10 msec (II), 102 ± 18 msec (III), and 185 ± 13 msec (IV). The peak amplitudes depended nonlinearly on intensity. This nonlinearity was most pronounced for Peaks I and II, whose amplitudes appeared to sample the initial part of the sigmoid-shaped curve modeling the strength of TMS-evoked neuronal activity. Although the response amplitude increased with stimulus intensity, scalp distributions of the potential were relatively similar for the four intensities. The results imply that TMS is able to evoke measurable brain activity at low stimulus intensities, probably significantly below 60% of MT. The shape of the response-stimulus intensity curve may be an indicator of the activation state of the brain. Hum. Brain Mapp. 21:154–164, 2004. © 2004 Wiley-Liss, Inc.