The response-variance-curve (RVC) method quantifies the variability of the individual epochs that constitute the average event related potential (ERP), providing complementary information to that offered by ERPs. Numerous studies have found that average ERP late components of an auditory "oddball" paradigm can differentiate schizophrenic patients from normal subjects. Our previous study of the RVC measure revealed significant differences between medicated and unmedicated schizophrenic patients in the maximum ERP variability from 190 to 240 ms. In the present study of unmedicated schizophrenic patients and normal control subjects, we examined the influence of intertarget intervals (generated by pseudorandom stimulus sequences in an auditory oddball paradigm) on the intratrial effects of ERP variability measured by the RVC. The ERPs of unmedicated schizophrenic patients were characterized by an instability in a latency window corresponding to the N200 component. The effect was particularly large at an intertarget interval of 7.8 s and was significantly reduced on either side of this intertarget interval. © 1995.