Microstate analysis of ERP and EEG topographies reveals stable patterns of brain activity. Little is known about the changes from one stable state to another. During such changes significant topographic instabilities are observed together with low ERP field power and high values of topographical dissimilarity. Although topographical dissimilarity indicates how stable a given topography is, it does not indicate the exact limits of stable and unstable intervals. In addition, it does not account for the influence of background noise on the degree of instability. In the current study, we introduce a measure that identifies unstable periods in topographical dynamics.We investigated the characteristics of stable and unstable periods in ERPs recorded in a target-detection paradigm where subjects counted target letters. Despite large variations among subjects, four main stable topographic periods and four unstable periods were found in the majority of ERPs. Stable periods were closel y related to the ERP waves (N1, P3a, P3b, N400). Scalar magnitudes of ERP vectors were significantly larger for the target condition than for the non-target condition in the time intervals containing the N1, P3b, and N400 stable periods. The longest unstable period occurred between the N1 and P3a waves (40 ms on average). The results suggest that our measure is a useful tool to identify and study the properties of unstable periods in ERPs.