The capability of predicting relapse in chronic alcoholism using quantitative EEG was investigated. For this purpose, 78 in-patients with alcoholism underwent EEG recordings (eyes closed) 7 days after the beginning of detoxification. Additionally, other clinical evaluations were carried out. After discharge from hospital, patients were regularly re-evaluated for the duration of 3 months in order to determine whether they relapsed or abstained from alcohol during this time. For classification of the two diagnostic subgroups (relapsers vs. abstainers), multivariate discriminant analysis as well as artificial neural network technology has been applied. Correct classification of patients' EEGs was achieved in 83-85% and thus outperformed classification with clinical variables considerably. Furthermore, artificial neural networks (ANN) improved classification results when compared with discriminant analysis. It was found that, in comparison to abstainers, relapsers had EEGs that were more desynchronized over frontal areas, which was interpreted as a functional disturbance of the prefrontal cortex.