The purpose of this study was to examine the association between intrapair birthweight discordance and fetal and neonatal mortality. This was a historical cohort study, using the Statistics Canada linked birth-infant death file. Twin births in Canada (excluding Ontario) for the period 1986-1997 were studied. Outcome measures were fetal and neonatal mortality. Among 59,034 twin births, 53% had 0 to 9% birthweight difference; 30% had 10 to 19% discordance; 11% had 20 to 29% discordance; and 6% had > or = 30% discordance. Maternal age, parity, sex discordance, and gestational age were important factors affecting birth weight discordance. Fetal death rates were 0.95, 1.26, 3.14, and 11.10%, respectively, in infants with a birthweight discordance of 0 to 9, 10 to 19, 20 to 29, and > or = 30%. Corresponding figures for neonatal death rates were 1.90, 2.32, 3.05, and 8.86%, respectively. Adjustment for important confounding variables including the actual birthweight and gestational age substantially reduced the birthweight discordance-related risk of fetal and neonatal mortality, but the results remained statistically significant and clinically important. The birthweight discordance-related risk of fetal and neonatal mortality was higher in smaller twins than in larger twins and higher in infants with gestational age > or = 37 weeks than those < 37 weeks. Birthweight discordance is an important risk factor for fetal and neonatal mortality, and the birthweight discordance-related risk of fetal and neonatal mortality is higher in smaller twins than in larger twins and higher in term twins than in preterm twins.