One of the main unsolved problems, and one which produces divergent results in clinical psychopharmacology, is that concerning the selection of patients and their diagnostic definition. An automated diagnostic procedure (PDA) was set up in order to classify each patient into one nosographic category on the basis of a cross-sectional examination of his mental state. Such diagnostic procedure appears particularly suitable for multicenter drug trials, since it gives a profile and a diagnostic definition of patients, assessed by investigators from different areas and with different cultural, and clinical backgrounds. In a multicenter trial (sulpiride versus haloperidol) PDA offered a chance to re-examine and analyze the characteristics of each patient and therefore to control the criteria followed for the sample selection in the various experimental settings. The agreement between clinician and computer diagnosis was 78.9%; this agreement rises to 85.5% if the computerlabelled schizo-affective syndromes are considered within the schizophrenic group. Moreover, and attempt has been made to relate psychopathological patterns to drug responses.