Objective To describe ethnic and socioeconomic variation in cause-specific infant mortality of preterm babies by gestational age at birth.

Design National birth cohort study.

Setting England and Wales 2006–2012.

Subjects Singleton live births at 24–36 completed weeks’ gestation (n=256 142).

Outcome measures Adjusted rate ratios for death in infancy by cause (three groups), within categories of gestational age at birth (24–27, 28–31, 32–36 weeks), by baby’s ethnicity (nine groups) or area deprivation score (Index of Multiple Deprivation quintiles).

Results Among 24–27 week births (5% of subjects; 47% of those who died in infancy), all minority ethnic groups had lower risk of immaturity-related death than White British, the lowest rate ratios being 0.63 (95% CI 0.49 to 0.80) for Black Caribbean, 0.74 (0.64 to 0.85) for Black African and 0.75 (0.60 to 0.94) for Indian. Among 32–36 week births, all minority groups had higher risk of death from congenital anomalies than White British, the highest rate ratios being 4.50 (3.78 to 5.37) for Pakistani, 2.89 (2.10 to 3.97) for Bangladeshi and 2.06 (1.59 to 2.68) for Black African; risks of death from congenital anomalies and combined rarer causes (infection, intrapartum conditions, SIDS and unclassified) increased with deprivation, the rate ratios comparing the most with the least deprived quintile being, respectively, 1.54 (1.22 to 1.93) and 2.05 (1.55 to 2.72). There was no evidence of socioeconomic variation in deaths from immaturity-related conditions.

Conclusions Gestation-specific preterm infant mortality shows contrasting ethnic patterns of death from immaturity-related conditions in extremely-preterm babies, and congenital anomalies in moderate/late-preterm babies. Socioeconomic variation derives from congenital anomalies and rarer causes in moderate/late-preterm babies. Future research should examine biological origins of extremely preterm birth.