Aims: Studies conducted in first-episode psychosis (FEP) samples avoid many biases. However, very few studies are based on epidemiological cohorts treated in specialized FEP services. The aim of this file audit study was to examine premorbid and baseline characteristics of a large epidemiological sample of FEP. Methods: File audit study of all patients admitted to the Early Psychosis Prevention and Intervention Centre between 1998 and 2000 using a specialized questionnaire. Results: There were 661 patient files included in the study. Premorbid evaluation revealed high rates of substance use disorder (74.1%), history of psychiatric disorder (47.5%), past traumatic events (82.7%) suicide attempts (14.3%) and family history of psychiatric illness (55.6%). Baseline characteristics revealed high intensity of illness (mean CGI 5.5), high prevalence of lack of insight (62%) and high rate of comorbidity (70%). Conclusion: High rates of traumatic events or episodes of mental illness before treatment for FEP must be considered when designing treatment approaches because a too narrow focus on positive psychotic symptoms will inevitably lead to incomplete treatment. Additionally, early intervention programmes need sufficient range of resources to address the multiple challenges presented by FEP patients such as high severity of illness, comorbidities and functional impairment. Finally, observation of an important degree of functional impairment despite short duration of untreated psychosis suggests that while early detection of FEP is a necessary step in early intervention, it may not be sufficient to improve functional recovery in psychosis and that efforts aimed at identifying people during the prodromal phase of psychotic disorders should be pursued.