The interrater reliability, confidence and ease of use of ICD-10 diagnostic criteria for research were assessed in the Republic of Korea as part of the field trials of World Health Organization collaborative study. A total of 279 patients were diagnosed according to the ICD-10 diagnostic criteria for research. Interrater reliability, calculated by kappa statistics, was found to be between 0.74 and 0.91 on 2-character categories and between 0.64 and 0.90 on 3-character categories except schizotypal disorder (F21). On the 4-character categories, the agreement in the affective states between raters was lower. The reasons are discussed based on cultural differences.