Objectives: Assessing the degree of agreement between the initial diagnosis of patient encounters made by referring institutions and the final diagnosis by a referred institution. Methods: Among 505 patients referred to the department of obstetrics and gynecology of one tertiary care hospital in 2003, 291 patients hospitalized with gynecologic problems were selected as study subjects in this study. Medical chart review was conducted to compare diagnostic codes given at the time of referral by the referring institutions and those given at discharge by the referred hospital. To assess the degree of accordance between the two diagnostic codes of the same patient encounter, we developed 5-point scoring system. The score of 5 indicates a perfect accordance of all 4 digits of two comparing diagnosis codes and 4 indicates accordance of the first 3 digits. The score of 3 is assigned if the two codes do not agree on the first 3 digits, but share the common disease area. The scores of 2, 1, and 0 are assigned for the cases of accordance on the first 2 digits and the first digit, and no accordance on the first digit, respectively. To identify factors predicting the degree of accordance, decision tree analysis using CART algorithm as data mining approach was carried out. Results: The average accordance score was 3.14. According to the decision tree analysis results, the type of specialty and health care institutions are primary factors that discriminate referrals with high accordance score from those with low score. Secondary factor was diagnosis group: the score was as high as 3.69 among referrals with diagnosis of benign tumor (diagnosis code: D25-D28) and carcinoma in situ (D00-D09), and 2.92 for other diagnosis groups. Tertiary factor was the type of patient`s occupation. Conclusions: Our study showed that there`s a substantial gap between referring and referred hospital on the diagnosis for the same encounter. Effective communication between the referring and referred hospitals may reduce the gap and consequently prevent medical waste resulting from duplicated procedures and tests.