OBJECTIVE To compare two dosing regimens for caffeine citrate for neonates born less than 30 weeks gestation in terms of development, temperament and behaviour. METHODS A multi-centre, randomised, controlled trial design was undertaken. A total of 287 infants with apnoea of prematurity or in the peri-extubation period were randomised to receive one of two dosage regimens (20 vs. 5 mg/kg/day). The main outcome measure was cognitive development at 1 year of age on the Griffiths Mental Development Scales. Secondary outcome measures included neonatal morbidity, death and disability, temperament at 1 year and behaviour at 2 years of age. RESULTS Data on the primary outcome were available for 190 survivors at 12 months corrected for prematurity. A significantly greater mean general quotient was found in the high-dose group (mean (standard deviation), 98.0 (13.8) vs. 93.6 (16.5), P = 0.048). On omission of two infants for whom cognitive assessment was not possible because of disability while the mean general quotient remained higher for infants in the high-dose group, this was no longer statistically significant (P= 0.075). There was a non-significant trend for benefit in the high-dose caffeine group for death or major disability, 15.4% versus 24.2%; relative risk 0.75 (95% confidence interval 0.49-1.14). No differences in the mean values between the two groups were shown for temperament and behaviour. CONCLUSIONS Caffeine citrate with a dosage regimen of 20 mg/kg/day did not result in adverse outcomes for development, temperament and behaviour. The borderline benefit in cognition with high-dose caffeine needs further investigation.