Transcranial magnetic stimulation (TMS) is a popular tool for mapping perceptual and cognitive processes in the human brain. It uses a magnetic field to stimulate the brain, modifying ongoing activity in neural tissue under the stimulating coil, producing an effect that has been likened to a “virtual lesion.” However, research into the functional basis of this effect, essential for the interpretation of findings, lags behind its application. Acutely, TMS may disable neuronal function, thereby interrupting ongoing neural processes. Alternatively, the effects of TMS have been attributed to an injection of “neural noise,” consistent with its immediate and effectively random depolarization of neurons. Here we apply an added-noise paradigm to test these alternatives. We delivered TMS to the visual cortex and measured its effect on a simple visual discrimination task, while concurrently manipulating the level of image noise in the visual stimulus itself. TMS increased thresholds overall; and increasing the amount of image noise systematically increased discrimination thresholds. However, these two effects were not independent. Rather, TMS interacted multiplicatively with the image noise, consistent with a reduction in the strength of the visual signal. Indeed, in this paradigm, there was no evidence that TMS independently added noise to the visual process. Thus, our findings indicate that the “virtual lesion” produced by TMS can take the form of a loss of signal strength which may reflect a momentary interruption to ongoing neural processing.