First-episode schizophrenia (FE-SZP) presents a diagnostic challenge because of symptomatic overlap between the various causes of psychosis. An early and accurate diagnosis is important for the implementation of appropriate treatment, for determining prognosis and for identifying research participants. In an effort to facilitate early diagnosis, we followed a group of first-episode psychosis patients with a presumptive diagnosis of schizophrenia who were subsequently diagnosed at 6-month follow-up with either schizophrenia (n=104) or other psychiatric diagnoses (n=19). The two groups-first-episode schizophrenia and first-episode non-schizophrenia-were compared on measures of demographics, symptoms, quality of life, premorbid adjustment and lateral dominance. Odds ratios were calculated for each variable and all significant variables were entered into a multivariate prediction model. The model showed that higher levels of anhedonia and hallucinations increased the odds of a final diagnosis of schizophrenia. This predictive model was validated in a smaller group of patients.