Background The EPICure 2 study shows increased survival (40–52%, p<0.0001) of admitted babies between 22 + 0 and 25 + 6 weeks gestation in England in 2006, compared with 1995.

Objective To explore reasons for increased survival by studying survival curves and changes in the population and clinical practice between 1995 and 2006.

Results Analysis of survival curves of admissions (1995: n  =  811; 2006: n  =  952) shows that the difference in survival is apparent shortly after birth; survival curves for babies still alive at 7 days are not significantly different at any gestation. Comparing demographic and clinical items known within 24 h of birth shows increased use of steroids (67–82%) and tocolysis (24–34%), fewer white (78–66%) and more Afro-Caribbean mothers (13–20%), more babies with heart rate greater than 100 bpm at 5 minutes (86–91%), less neonatal hypothermia (40–16%) and more surfactant use (85–97%), all p⩽0.001. There was a non-significant trend towards a lower CRIB 1 score at all gestations. Logistic regression analysis using these and other factors with the outcomes: death, severely abnormal head scan, treated retinopathy of prematurity and oxygen supplementation at expected delivery date shows similar patterns of association in the two cohorts.

Conclusions The increase in survival for babies less than 26 weeks seen in EPICure 2 is confined to the first week. This is associated with evidence of babies being in better condition around the time of birth.