This paper describes the long-term course of 76 patients who had been consecutively admitted to the Department of Child and Adolescent Psychiatry, Philipps University, between 1920 and 1961 with a suspected diagnosis of childhood-onset schizophrenia. By means of a consensus analysis of available data in accordance with ICD-10 criteria, the diagnosis of schizophrenia was confirmed in only 50% of the original sample (n = 38, childhood-onset schizophrenia group); whereas the rest of the sample were allotted other diagnoses (n = 38, non-schizophrenia group). A follow-up investigation was conducted, interviewing all available patients, if possible, or their first-degree relatives or doctors. In the childhood-onset schizophrenia group, age at onset (mean +/- S.D.) was 12.7 +/- 2.5 (range 5-14) years and age at follow-up was 55.0 +/- 4.8 (range 42-62) years. The outcome of this group was poor. According to the Global Assessment Scale (GAS), only 16% had a good (GAS score 71-100) and 24% had a moderate (GAS score 41-70) outcome. In the 16 childhood-onset schizophrenia patients who could be personally investigated at follow-up, 10 (62.5%) displayed severe or moderate depressive symptoms according to the BPRS depressive score. The death rate (including suicide) was significantly higher in the schizophrenia group (n = 15; 39.5%) than in the non-schizophrenia group (n = 7; 18.4%). A comparison of the life-time diagnoses of the total sample (n = 76) at follow-up with the ICD-10 diagnoses made retrospectively revealed a diagnostic stability in 69 (91%) and a change of diagnosis in 7 (9%) cases, among them 4 who were originally diagnosed as having childhood-onset schizophrenia.