In the quest for neurophysiological biomarkers that uniquely characterize schizophrenia subjects, auditory evoked potentials (EPs) have been extensively used during the past several decades. Typically, EPs are estimated using ensemble averaging to obtain robust components. Averaging, however, eliminates all temporal variability of the recorded signals and, therefore, hampers the study of the brain temporal dynamics underlying the generation of EP components. In this chapter, we present a methodology for analyzing EPs on a single-trial basis using an iterative independent component analysis procedure. The method is capable of identifying and measuring the amplitude, latency, and overall morphology of individual EP components in single trials and, as such, permits the study of phase characteristics among single trials while preserving known features of the average EPs. Recordings from schizophrenia patients and normal controls demonstrate that activity phase synchronization plays a crucial role in EP generation and explains the sensory gating deficits observed in schizophrenia subjects. Furthermore, the findings from this method are very robust across recordings from different labs and experimental protocols and can be used to separate schizophrenia patients from normal controls with 100 % classification accuracy.