A true science of psychiatry is in need of a better taxonomy. Traditional diagnostic approaches, including the recently developed DSM-III criteria, are not objective. Arguments are offered in favour of the polythetic classification strategy, based on the principles of numerical taxonomy. The hierarchical complexity of biological systems, including man, should be taken into account in applying this approach. Polythetic classifications which are based on variables limited to a single level of biological organization may offer advantages to 'true' polythetic classifications, based on as many levels as possible. Cerebral information processing represents an intermediate level of analysis which may serve as a bridge between the research fields of biological and social psychiatry. This level is particularly promising in the search for an alternative to the schizophrenia concept.