This study describes the evolution in fetal and neonatal mortality rates among twin pairs born in 22 hospitals located in the eastern regions of the province of Quebec in 1976-1978 (n = 776 pairs) and 1982-1985 (n = 712 pairs). It also assesses the contribution of maternal factors, obstetrical care and characteristics of twins in the variation of the risk of death over time. The fetal mortality rate did not improve from 1976-1978 (22.6 per 1000) to 1982-1985 (28.1 per 1000). However, the neonatal mortality rate declined from 44.7 to 34.7 per 1000 liveborn first twins and from 56.8 to 36.1 per 1000 liveborn second twins. For first twins as for second twins, birthweight-specific neonatal mortality rates decreased within birthweight categories under 2500 g. In the second period, 96.9% of twin pregnancies were detected before confinement compared to 59.6% in the earlier period. The proportion of twins delivered by obstetricians, the percentage of twin births occurring in ultraspecialized perinatal units and the frequency of caesarean sections increased markedly. The proportion of preterm births increased over time (34.5% vs 43.1%) whereas the percentage of low birthweight twins decreased but not significantly (54.3% 51.6%). In this study, changes in maternal age, parity, educational level, sex of pairs, qualification of the physician, and level of care available at the hospital of birth, did not account for the decrease in neonatal mortality rates among twins. The increase in the frequency of caesarean sections seemed to explain only a small proportion of the decrease in the neonatal mortality rate among second twins. In the second as well as in the first period, the neonatal mortality rate for twins was six times higher than that for singletons.