Background Barker's fetal programming hypothesis suggests that disproportionate size at birth may have a lifelong impact on one's health. However, the literature on birth length is considerably more sparse compared with birthweight. We, therefore, examined the relationship between birth length and hospitalisation early in life among Japanese children. Methods We used the nationwide Longitudinal Survey of Babies in 21st Century and restricted the study subjects to full-term singleton babies (n=44057). We estimated the effects of birth length and birthweight on the risk of hospitalisation using log linear regression models. We controlled for a set of neonatal and maternal factors. Results Birth length was associated with the chance of hospitalisation due to all causes between 6 and 18 months of age. In addition, the association was stronger than that with birthweight. Adjusted risk ratios showed that the relationship between birth length and hospitalisation was U-shaped: 1.16 [95% confidence intervals, 1.08, 1.25] at 30-48cm, 1 [Reference] at 49cm, 1.13 [1.04, 1.22] at 50cm, and 1.11 [1.02, 1.20] at 51-60cm. Short babies with low or high weight, as well as long babies with low weight, seem to be at increased risk of hospitalisation. Conclusions We found a U-shaped relationship between birth length and risk of hospitalisation due to all causes during the period from 6 to 18 months.