This study investigated the temporal relation between two early mechanisms of auditory information processing: the segregation of the auditory input into streams and the automatic deviance detection indicated by the mismatch negativity (MMN). To address this question the processing of successive deviant and standard tones within streaming and nonstreaming conditions was analyzed. In the streaming condition the amplitude reduction of MMN elicited by the second of two successive deviants was found to be smaller for successive deviants presented in different than in same streams. No corresponding MMN differences were obtained in a nonstreaming condition. These results demonstrate that stream segregation precedes deviance detection. Moreover, modulations of the N1 amplitudes elicited by successive standard tones in the streaming condition demonstrate that not only deviance-related processing but even initial sound processing is affected by streaming.