Objective : To evaluate the perinatal outcomes of twin pregnancies with birth weight discordancy and to determine factors affecting on the outcomes. Methods : We studied 367 pairs of twin delivered after 28 weeks of gestation at the Ajou University Hospital between June 1994 and June 2000. Twins were stratified into three groups according to the percent difference of birth weight. Birth weight difference less than 15%(concordant) was classified into control group(268 pairs), between 15% and 25% into group 1(72 pairs), and more than 25% into group 2(27 pairs). Perinatal outcomes of each group were assessed retrospectively. Results : In smaller twins, mean birth weight(control vs group 1 vs. group 2 : 2274.3±424.9 gm vs. 2012.9±303.2 gm vs. 1635.2±440.8 gm, p<0.05), duration of neonatal intensive care unit admission(7.6±11.5 days vs. 11.6±10.8 days vs. 18.6±14.4 days, p<0.05), and the frequency of neonatal jaundice(8.6% vs. 12.5% vs. 29.6%, p<0.05), the frequency of small for gestational age infant(8.2% vs. 40.3% vs. 74.1%, p<0.05), and perinatal mortality(2.6% vs. 2.8% vs. 11.1%, p<0.05) in group 2 showed statistically significant difference from control group. No difference was found in larger twins. In multiple logistic regression analysis, independent prognostic factors of discordant twin were gestational age and birth weight. Conclusion : In twin pregnancies with birth weight discordancy, larger twins showed no difference in perinatal outcomes but smaller twins with birth weight discordancy more than 25% showed significantly higher perinatal mortality and morbidity. However independent prognostic factor was not discordancy itself but gestational age and birth weight.