A Dutch translation of the PSE (9th edition) was tested for inter-rater reliability in one hundred newly admitted patients with a broad variety of symptoms. Twenty tour of them were re-interviewed in order to assess test-retest reliability. Seven raters took part in the study. The reliabilities of symptoms, syndromes, and CATEGO-classes were comparable to those reported in the literature. Ratings based on direct observation showed low reliability in contrast to those based on verbal information. The influences of the patients' diagnostic characteristics and of rater variables such as training, experience , and role on inter-rater reliability turned out to be smalt. The intluence of diagnostic characteristics on the test-retest reliability, however, was considerable, i.e. non-depressives were rated and classified significantly less reliable than depressives.