In the following, the findings from a field study on the reliability of ICD-10 for classifying psychiatric disorders in children and adolescents are presented. In the main study there were 33 raters from 9 different facilities; each rater assessed 29 case studies independently. The evaluations of the 29 case studies independently. The evaluations of the 29 case histories yielded 957 ratings. When 4-digit ICD-10 codes were used, the most common main diagnosis was coded by an average of 50.3% (ICD-9: 47.3%). When 3-digit ICD-10 codes were used, the figure was 60.6% (ICD-9: 66.5%), that is, as expected, agreement was clearly better. When alternative diagnoses were considered, agreement was substantially better; the same was true for intensive schooling of the raters (additional study). Some diagnoses were rather difficult to differentiate from each other. Among these were the mixed disorders of conduct and emotions (F92). The main problem was differentiating between this disorder and the categories conduct disorder (F91), emotional disorder (F93), affective disorders (F30-F39), and neurotic disorders (F40-F49). There were also problems classifying age-related depressive disorders which in ICD-9 could be coded under 313.1, probably in part because this category does no longer exist in ICD-10. These disorders have now to be coded as depressive disorders (F31.1, F33.1 or F34.1). The diagnostic descriptions of these categories are based mainly on the symptoms typical in adults. No facility-related tendencies to code in a particular way are evident.