Objective To examine how cognitive ability is related to gestational age and the extent to which observed differences could be explained by socio-economic confounding.

Methods The Millennium Cohort Study collected data on 18 818 children at 9 months and 3, 5 and 7 years. Cognitive development was assessed using Bracken School Readiness Assessment at age 3, British Ability Scales II at ages 3, 5 and 7 and progress in mathematics at 7 years. Z-scores were analysed by linear regression with adjustment for confounders.

Results Children born at <32 weeks gestation scored consistently poorer on all scales while the results were less consistent for children born at later gestational ages. We found the strongest association between gestational age and cognitive ability on the scales representing spatial and mathematical ability with z-scores for the children born at 34–36 weeks and 37–38 weeks that were between 0.1 and 0.2 SD lower than children born at 39+ weeks. The Bracken School Readiness Assessment at 3 years also showed statistical significant differences for the later gestational age groups. After adjustment for maternal education and other factors, the differences between the term and preterm groups were attenuated, but the overall pattern remained.

Conclusions These results suggest that cognitive ability is related to the entire range of gestational age, including late preterm and early term birth. However, the differences were small for later gestational age groups and were partly explained by socio-economic effects as a higher proportion of children in these groups live in socially disadvantaged circumstances.