Since Kraepelin's first description of dementia praecox in 1889 many data and theories have been published on the onset and course of schizophrenia. Until now studies on these topics had to rely on first admission data and on the subsequent course of the disease. However, first hospitalisation is preceded by a wide variety of patterns and duration of the early course. Items taken from the pre-admission phase of the disease are often incorrectly used as premorbid characteristics, understandably preceding the subsequent course and outcome of schizophrenia with high predictive power. In relation to our interest to study the beginning of schizophrenia, systematically, paying special attention to the age and gender distribution of true onset and the symptomatology and pattern of the early and later course, we developed an 'Interview for the Retrospective Assessment of the Onset of Schizophrenia (IRAOS)'. It allows an objective, reliable, and valid assessment of the symptoms, psychological impairments, demographic and social characteristics as well as the referring points in time of the early course of psychosis. The instrument is administered as a semi-structured interview with both the patient and a key informant. The obtained information is extended by a systematic examination of the clinician's case notes. Some results derived from an ongoing study on age and gender differences in onset and patterns of early course are added to demonstrate the use of the instrument.