Objective To analyse the effect of different preterm birth categories on behavioural outcomes.

Design Prospective cohort design.

Settings The Japanese Longitudinal Survey of Babies in the 21st Century was used for the study.

Subjects Neonates born in 2001 in Japan (n=34 163) and followed up at the age of 8 years. Post-term births were excluded and the sample was divided into four groups according to gestational age: <34 weeks, 34–36 weeks, 37–38 weeks and 39–41 weeks.

Outcome measures Seven behavioural outcomes were examined: three attention problems and four delinquent/aggressive behaviours.

Results Logistic regression was used to estimate adjusted ORs and 95% CIs for each outcome. Positive associations were found between preterm birth (<37 weeks) and adverse behavioural outcomes compared with full-term birth (39–41 weeks). For attention problems, the OR for ‘inability to wait his/her turn’ was increased by 1.72 (95% CI 1.22 to 2.43) for gestational age <34 weeks and by 1.28 (95% CI 1.03 to 1.59) for 34–36 weeks’ gestation. For delinquent/aggressive behaviour, the OR for the outcome ‘destroying books/toys’ was increased by 1.46 (95% CI 1.07 to 1.99) for gestational age <34 weeks, while the outcome ‘disturbance in public’ was increased by 1.20 (95% CI 1.04 to 1.38) for 34–36 weeks’ gestation.

Conclusions Children who are born <34 weeks, but also between 34 and 36 weeks, are at increased risk of behavioural problems related to attention and delinquent/aggressive behaviour when 8 years old.