BACKGROUND: The prevalence of multiple births has recently increased. However, the association between gestational age and long-term morbidity among twins remains unclear. AIMS: To examine the association of gestational age with child health and neurological development in early childhood among twins. STUDY DESIGN: Population-based longitudinal study. SUBJECTS: We included 947 children from 479 pairs of twins with information on gestational age. OUTCOME MEASURES: Hospitalization was used as an indicator of physical health, and responses to questions about age-appropriate behaviors were used as an indicator of neurobehavioral development. We conducted binomial log-linear regression analyses, controlling for both child and maternal variables in the model. We accounted for correlations within the pairs with generalized estimating equations. RESULTS: The early term group (i.e., 37 to 38 weeks of gestation) had a lower risk of poor child health and unfavorable neurodevelopment compared with the full term group (≥39 weeks of gestation) and preterm group (<37 weeks of gestation). Compared with the early term group, the adjusted risk ratios for hospitalization for all causes during the period from 7 to 18 months of age was 2.2 (95% confidence interval: 1.3-3.8) for very preterm children (<32 weeks of gestation), 1.1 (0.8-1.6) for moderately and late preterm children (32 to 36 weeks of gestation), and 1.8 (1.0-3.2) for full term children. CONCLUSION: We observed a U-shaped association of gestational age with child health and neurodevelopment. The early term group had the lowest risk of poor outcomes among twins.