Background The drainage, irrigation and fibrinolytic therapy (DRIFT) trial, conducted in 2003–6, showed a reduced rate of death or severe disability at 2 years in the DRIFT compared with the standard treatment group, among preterm infants with intraventricular haemorrhage (IVH) and post-haemorrhagic ventricular dilatation. Objectives To compare cognitive function, visual and sensorimotor ability, emotional well-being, use of specialist health/rehabilitative and educational services, neuroimaging, and economic costs and benefits at school age. Design Ten-year follow-up of a randomised controlled trial. Setting Neonatal intensive care units (Bristol, Katowice, Glasgow and Bergen). Participants Fifty-two of the original 77 infants randomised. Interventions DRIFT or standard therapy (cerebrospinal fluid tapping). Main outcome measures Primary – cognitive disability. Secondary – vision; sensorimotor disability; emotional/behavioural function; education; neurosurgical sequelae on magnetic resonance imaging; preference-based measures of health-related quality of life; costs of neonatal treatment and of subsequent health care in childhood; health and social care costs and impact on family at age 10 years; and a decision analysis model to estimate the cost-effectiveness of DRIFT compared with standard treatment up to the age of 18 years. Results By 10 years of age, 12 children had died and 13 were either lost to follow-up or had declined to participate. A total of 52 children were assessed at 10 years of age (DRIFT, n = 28; standard treatment, n = 24). Imbalances in gender and birthweight favoured the standard treatment group. The unadjusted mean cognitive quotient (CQ) score was 69.3 points [standard deviation (SD) 30.1 points] in the DRIFT group compared with 53.7 points (SD 35.7 points) in the standard treatment group, a difference of 15.7 points, 95% confidence interval (CI) –2.9 to 34.2 points; p = 0.096. After adjusting for the prespecified covariates (gender, birthweight and grade of IVH), this evidence strengthened: children who received DRIFT had a CQ advantage of 23.5 points (p = 0.009). The binary outcome, alive without severe cognitive disability, gave strong evidence that DRIFT improved cognition [unadjusted odds ratio (OR) 3.6 (95% CI 1.2 to 11.0; p = 0.026) and adjusted OR 10.0 (95% CI 2.1 to 46.7; p = 0.004)]; the number needed to treat was three. No significant differences were found in any secondary outcomes. There was weak evidence that DRIFT reduced special school attendance (adjusted OR 0.27, 95% CI 0.07 to 1.05; p = 0.059). The neonatal stay (unadjusted mean difference £6556, 95% CI –£11,161 to £24,273) and subsequent hospital care (£3413, 95% CI –£12,408 to £19,234) costs were higher in the DRIFT arm, but the wide CIs included zero. The decision analysis model indicated that DRIFT has the potential to be cost-effective at 18 years of age. The incremental cost-effectiveness ratio (£15,621 per quality-adjusted life-year) was below the National Institute for Health and Care Excellence threshold. The cost-effectiveness results were sensitive to adjustment for birthweight and gender. Limitations The main limitations are the sample size of the trial and that important characteristics were unbalanced at baseline and at the 10-year follow-up. Although the analyses conducted here were prespecified in the analysis plan, they had not been prespecified in the original trial registration. Conclusions DRIFT improves cognitive function when taking into account birthweight, grade of IVH and gender. DRIFT is probably effective and, given the reduction in the need for special education, has the potential to be cost-effective as well. A future UK multicentre trial is required to assess efficacy and safety of DRIFT when delivered across multiple sites. Trial registration Current Controlled Trials ISRCTN80286058. Funding This project was funded by the NIHR Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 23, No. 4. See the NIHR Journals Library website for further project information. The DRIFT trial and 2-year follow-up was funded by Cerebra and the James and Grace Anderson Trust.