Our starting hypothesis was that schizophrenic symptomatology changes over time. This hypothesis explains conflicting reports on schizophrenic symptom structures as a consequence of different durations of illness in the samples studied to date. Therefore a sample of 258 schizophrenic in-patients (with ICD-10 diagnoses F20) was categorized according to illness duration. A factor analysis was performed on the 8 items of the Manchester Scale for three subgroups (duration < 10 years, 10> or =20 years and > or = 20 years). For those patients whose illness duration was less than 10 years, 'formal thought disorder' was not related to any other mental state, whereas for those whose duration was 10 years or longer, it was correlated with 'negative symptoms'. In the < 10 years group, 'anxiety and depression syndrome' and 'positive symptoms' formed one complex, but these symptoms were separated into two distinct syndromes in the > or = 20 years group. Thus we were able to demonstrate that the classification of symptoms changes with increasing duration of illness.