Background:  Twin pregnancies have an elevated risk of adverse outcomes, particularly preterm twins.
Aims:  Describe the distribution of twin deliveries by hospital level, the associated perinatal and maternal morbidity, and determine predictors of perinatal morbidity and urgent transfer to a neonatal intensive care unit.
Methods:  Longitudinally linked New South Wales delivery and hospital records for the years 2001–2005 were used to identify perinatal and maternal morbidity/mortality in twin pregnancies. Regression analysis was used to examine predictive factors, including birth hospital volume.
Results:  At ≤ 32 weeks, 88.1% of twins were delivered in tertiary referral hospitals. By 34–35 weeks, only 39.7% of twins were delivered in tertiary units. Gestational age was the primary predictor of perinatal morbidity/mortality. Perinatal morbidity/mortality and maternal morbidity were lowest for deliveries at 38 weeks. There was no evidence that planned caesarean section at ≤ 38 weeks was protective against perinatal morbidity/mortality. There was an increased risk of perinatal morbidity/mortality (odds ratio (OR) = 2.22) for twins delivered at 33–35 weeks gestation at hospitals with < 500 deliveries per annum, and an increased risk of urgent neonatal transfer (OR = 2.06). Twin pairs for whom there was a ≥ 20% discordance in birthweight had an increased risk of morbidity/mortality at 36–38 weeks (OR = 1.79). Conclusions:  Both infant and maternal morbidity increase from 39 weeks gestation. Delivery of twins before 36 weeks at smaller hospitals (< 500 deliveries per annum) should be avoided. A twin pregnancy where there is a ≥ 20% difference in estimated fetal weights should be considered for referral to a tertiary obstetric unit.