Early identification of schizophrenia in patients with a first episode of psychosis (FEP) may help to avoid inappropriate treatment and may enhance long-term outcome by addressing issues such as family network, treatment adherence and functional and symptomatic outcome. It was the aim of the study to determine baseline variables that significantly predicted a diagnosis of schizophrenia in patients with FEP. The sample consisted of 133 FEP patients hospitalized for at least 6 weeks, in whom a DSM-IV diagnosis was confirmed after 1 year follow-up. Patients were divided into two groups, those with a diagnosis of schizophrenia (Schizophrenia group, n=63; 47.8%), and those with other psychosis, who were grouped under Non-Schizophrenic Psychosis (NSP, n=70; 52.2%). Sociodemographic (marital status, educational level) and clinical variables were recorded for each patient. Substance use (alcohol, cannabis and cocaine) did not statistically differ between the two groups. Absence of characteristics defined as criteria for good prognosis, lack of > or = 20% improvement in the total Positive and Negative Syndrome Scale score at 6 weeks, and a poor premorbid adjustment as determined by the Premorbid Adjustment Scale score significantly predicted the presence of schizophrenia. The regression model including these three variables achieved a predictive value of 76.3%, with a sensitivity of 74.6% and a specificity of 77.9%.