Introduction: MMN, an electrophysiological measure of auditory working memory, is usually recorded as the difference in the event-related potential (ERP) elicited by a rare deviant and a common standard sound. The amplitude is usually reduced in patients with schizophrenia (refs below). Here we looked at the response in the extreme and most simple case of the deviant being always a novel, different tone on every presentation. We compared the novelty-MMN with clinical symptoms expressed - both measures being made soon after admission and again 2-3 months later just before discharge. Methods: We compared 20 patients (mean age 26y) with a first, second or third episode of schizophrenia on admission with 21 healthy controls (mean age 26y) and were able to repeat the measures with 12 patients at discharge and with 15 controls. An early MMN component (80-140 ms), a later component (140-300 ms) and the P3a were recorded and topography examined after min-max norming from 19 sites. Symptoms were assessed with the positive and negative syndrome scale (PANSS). Results: a) Novelty-elicited MMN was not significantly reduced on admission (Fig. 1). b) The early component remained unaltered, but the amplitude of the later component decreased significantly during inpatient treatment. While the decrease appeared significant over the left hemisphere in the raw data, the lateral difference was lost after normalizing the data (Fig. 2). c) Improved positive symptom ratings were associated with increases of the early component latency, but decreases of the late component latency. d) P3a at Fz showed an increased latency between sessions in the patients but there were no group differences in amplitude. Conclusions: Our results are partially consistent with two other studies using a conventional MMN that showed a lack of MMN normalization where symptoms improved (Schall et al., 1998; Umbricht et al., 1998) - in the present study MMN deteriorated. While trait features have been attributed to conventional MMN reductions in schizophrenia, our results suggest that if novelty responses are impaired in patients with schizophrenia then the differences may be sensitive to state.