Interpretation of neurophysiological differences between control and patient groups on the basis of scalp-recorded event-related brain potentials (ERPs), although common and promising, is often complicated in the absence of information on the distinct neural generators contributing to the ERP, particularly information regarding individual differences in the generators. For example, while sensory gating differences frequently observed in patients with schizophrenia in the P50 paired-click gating paradigm are typically interpreted as reflecting group differences in generator source strength, differences in the latency and/or orientation of P50 generators may also account for observed group differences. The present study examined how variability in source strength, amplitude, or orientation affects the P50 component of the scalp-recorded ERP. In Experiment 1, simulations examined the effect of changes in source strength, orientation, or latency in superior temporal gyrus (STG) dipoles on P50 recorded at Cz. In Experiment 2, within- and between-subject variability in left and right M50 STG dipole source strength, latency, and orientation was examined in 19 subjects. Given the frequently reported differences in left and right STG anatomy and function, substantial inter-subject and inter-hemispheric variability in these parameters were expected, with important consequences for how P50 at Cz reflects brain activity from relevant generators. In Experiment 1, simulated P50 responses were computed from hypothetical left- and right-hemisphere STG generators, with latency, amplitude, and orientation of the generators varied systematically. In Experiment 2, electroencephalographic (EEG) and magnetoencephalographic (MEG) data were collected from 19 subjects. Generators were modeled from the MEG data to assess and illustrate the generator variability evaluated parametrically in Experiment 1. In Experiment 1, realistic amounts of variability in generator latency, amplitude, and orientation produced ERPs in which P50 scoring was compromised and interpretation complicated. In Experiment 2, significant within and between subject variability was observed in the left and right hemisphere STG M50 sources. Given the variability in M50 source strength, orientation, and amplitude observed here in nonpatient subjects, future studies should examine whether group differences in P50 gating ratios typically observed for patient vs. control groups are specific to a particular hemisphere, as well as whether the group differences are due to differences in dipole source strength, latency, orientation, or a combination of these parameters. Present analyses focused on P50/M50 merely as an example of the broader need to evaluate scalp phenomena in light of underlying generators. The development and widespread use of EEG/MEG source localization methods will greatly enhance the interpretation and value of EEG/MEG data.