The subjects were 34 acutely ill in-patients who met the RDC criteria of schizophrenic psychosis, and 4 EEGs were recorded from each patient before, 2 h and 24 h after oral intake of a single dose of 150 mg perazine, and on the 28th day of the neuroleptic treatment period. As a criterion of clinical response a decrease of at least 66% in the schizophrenia-specific sum score of the Brief Psychiatric Rating Scale on day 28 relative to the baseline value was decided upon. The EEGs were assessed using a newly developed procedure which takes into consideration 4 derivations simultaneously. As we tried to search out EEG variables with predictive value the statistical data analysis underlying our findings should largely by regarded as exploratory. Independent of day, responders (R) showed a tendency towards more low voltage desynchronized epochs (non-A stage) than non-responders (NR). Thus, R exhibited a higher degree of dynamic variability or a broader range of control of the spontaneous vigilance fluctuation (dynamic lability) than NR (dynamic rigidity). Furthermore, R and NR differed with respect to their time-dependent changes of non-A epoch frequencies before medication. While R showed a monotonous increase which is typical for normals, NR did not. Because of considerable inter-individual variability these group differences could not be used for individual prediction of the therapy response. By means of a qualitative data analysis R could be distinguished from NR with regard to various test dose-induced changes of the topographical distribution of absolute alpha power. All the group differentiating variables showed a time course of the same kind: R showed a prompt and ample deflection and the same recovery of baseline; NR, in contrast, showed no significant deflections at all. These findings are in line with the results concerning the dynamics of vigilance and certain claims of earlier authors according to which EEG changeability should be decisive for therapeutic outcome.(ABSTRACT TRUNCATED AT 400 WORDS)