The present paper shows that the short-term prognosis of patients with paranoid disorders hospitalized nowadays is poor for the majority. The best prognosis is attached to patients with affective or other not specified psychotic disorders. Socio-vocational variables at the time of first admission contain more predictive value than clinical variables. The term reactive psychosis as a term with good prognosis is questioned and the findings of a mortality study are presented in support of this view when concerned with hospital populations.