The objective of this study was to determine the effects of birth weight and gestational age on twin vs. singleton mortality. Population-based analysis of live births, fetal deaths, and infant deaths by plurality in the United States from 1983 to 1986 was conducted. Seven mortality rates and relative risks (RRs) of twin vs. singleton mortality were calculated by birth weight, gestational age, and combined birth weight and gestational age. The mortality rates included fetal, perinatal, early neonatal, late neonatal, neonatal, postneonatal, and infant. Twins had 3–4 times the RRs of mortality compared to singletons, ranging from a RR of 2.71 for postneonatal mortality to a RR of 3.73 for late neonatal mortality. Generally, for birth weights of 2,800 g or less and gestational ages of 38 weeks or less, twins had lower combined birth weight and gestational age mortality rates and lower RRs. Between 1,900 and 2,799 g, mortality rates decreased then increased with advancing gestation between 31 and 42 weeks both more severely and consistently for twins than for singletons. In conclusion, twins have lower birth weight and gestational age-specific mortality rates and RRs than singletons below 2,800 g and 39 weeks. The “U”-shaped pattern of mortality beyond 38 weeks gestation, particularly for twins with birth weights below 2,500 g, reflects the combined influence of growth retardation and advancing gestation on mortality. The lowest mortalityfor twins is achieved at birth weights of 2,500-2,799gat35-38 weeks gestation. Only 1 in 7 twins is born within this “ideal window.” Efforts at reducing twin mortality should be directed toward reducing intrauterine growth retardation and achieving optimal timing for delivery.