The Brief Psychiatric Rating Scale (BPRS) is expected to provide a valuable instrument for routine evaluation of the patient's status and correct interpretation of concentrations measured during therapeutic monitoring of antipsychotic drugs. Inpatient files were searched for information relative to acute episodes and BPRS items were scored every 10 days until discharge from the hospital according to a retrospective scheme. Eighty patients were included, who met DSM III criteria for schizophrenia and bipolar disorder. Clusters of symptoms that tend to appear together were identified through principal components analysis and are in keeping with factors described previously. Differences with respect to diagnosis further suggest that such clusters convey clinically relevant information. The time course of haloperidol action was also investigated: an average 51 % improvement was observed over the first 10 days, while the proportion of patients showing a 50% or larger BPRS decrease at the time of discharge reached 90%. The prominent contribution of changes in positive, florid symptoms to global improvement is also described. Finally, association of clinical improvement with a shorter period of hospitalization is shown. These results indicate that the BPRS represents a level of abstraction compatible with the way clinicians communicate and that its introduction into routine practice may allow for better description of the course of illness