This study examined whether assessment data from administration of the extended version of the Brief Psychiatric Rating Scale (BPRS-E) to state hospital patients within 72 hours of their admission could be used to predict length of hospital stay. BPRS-E data for 222 first-admission patients, for whom the mean length of stay was 118.4+/-88.6 days, were factor analyzed, yielding a model with four factors: patient's resistance to treatment, positive symptoms, mood, and negative symptoms. Discriminant analysis showed that the negative symptoms factor (blunted affect, emotional withdrawal, motor retardation, self-neglect, and disorientation) correctly predicted whether length of stay would exceed the mean in 94 percent of cases. The findings suggest that the severity of negative symptoms can be a useful predictor of length of stay among patients with severe and persistent mental illness.