Aim: This paper reports the rationale, methodology and baseline characteristics of a large long-term follow-up study of first-episode psychosis from a geographically defined catchment area. Method: A total of 723 first-episode psychosis patients were recruited from a specialized early psychosis service between 1989 and 2001 and prospectively followed up at a median of 7.4 years after initial presentation. Participants' baseline demographic, clinical and functional characteristics are described. Sampling bias at study recruitment was assessed by comparison with a more complete sample of Early Psychosis Prevention and Intervention Centre (EPPIC) cases rated directly from the medical records. Results: At baseline, 57% of the sample were diagnosed with schizophrenia or schizophreniform disorder, whereas the full range of psychotic disorders was represented. Statistical analysis confirmed that the sample recruited was representative of total EPPIC-treated incident cases. Conclusions: The EPPIC long-term follow-up study is a large and epidemiologically representative first-episode psychosis cohort that has been subsequently prospectively followed up over a long period. Such a sample provides a rare opportunity to study the course and outcome of psychotic disorders.