We propose and demonstrate a novel method for analyzing human EEG at a single subject and single-trial level. We focus here on the analysis of data from an auditory object recognition experiment. The analysis is based on the topographic information that can be extracted in response to each stimulus. Temporally structured periods were statistically identified wherein a given topography predominated without any prior information about the temporal behavior and at single-subject level. Within these periods of stability, we were able to statistically identify the time periods over which responses to different object categories differed. In addition to providing novel methods for EEG analysis, these results show intriguing evidence that Event Related Potentials (ERPs) are reliably observable at a single-trial level when examined topographically.