We have recently developed an iterative procedure based on independent component analysis to estimate single trial evoked potentials and used the procedure to analyze data obtained from normal subjects in response to a pure tone, with a whole-head 256-channel recording system. Our results show that new methodology can effectively remove most well-defined artifacts and background activity, and also provide improved estimates of specific components of the evoked potential. With respect to the N100 component, the processed single trials could be separated into two groups, one with trials having the same phase as the average response and another one with trials having the opposite phase, while the number of trials in each group had a very consistent spatial distribution across all subjects. Finally, the sources of the N100 component were primarily due to in-phase responses, while the out-of-phase responses had an antagonistic effect.