Commentary on: Hintz S, Barnes P, Bulas D, Slovis T, Finer N, Wrage L, et al. Neuroimaging and neurodevelopmental outcome in extremely preterm infants. Pediatrics 2015; 135: e32–42. Neurodevelopmental outcomes are a crucial part of prognosis and are pivotal in ongoing conversations between neonatologists and parents 1, 2. Recent studies show greater predictive value in near-term cranial ultrasound (CUS) and MRI versus traditional CUS performed in the first few weeks of life 1-3. MRI detects white matter injury more accurately while US detects intraventricular haemorrhages, ventriculomegaly and cystic periventricular leukomalacia 2, 4, 5. However, studies have found former preterm infants with developmental impairments despite normal NICU ultrasounds, and, importantly, others have reported transient findings of cystic periventricular leukomalacia (resolving by near-term) are associated with developmental impairments 2, 6, 7. The NEURO prospective study by Hintz et al. 8 analysed the predictive ability of early CUS, near-term CUS and near-term MRI in a cohort of 480 extremely preterm neonates, born at 24–27 weeks. Negative outcomes were defined as neurodevelopmental impairment, significant gross motor impairment or death. The authors provided multivariable modelling to consider perinatal and neonatal demographic and clinical risk factors. Results were presented as OR and AUC of ROC curves. Further statistical calculations of Tables 2–5 show sensitivity and specificity of 0.27 and 0.92, respectively, for the association between major findings in early CUS and negative outcomes of neurodevelopmental impairment or death, compared to sensitivity of 0.28 and specificity of 0.97 for major findings in late CUS. The calculated sensitivity and specificity of MRI WMA are similar to that reported by Woodward et al. 4 in 2006 (92% vs 84% and 23% vs 34%, respectively), demonstrating that MRI has potential for higher sensitivity than CUS. The major findings of the NEURO study were that early CUS is not predictive of future neurodevelopmental outcomes, despite taking into account perinatal and neonatal demographic and clinical risk factors. The ability to predict outcomes is only mildly enhanced by the addition of MRI to late CUS. Therefore, in environments where MRI is not easily available, late CUS may be sufficient in predicting prognosis. The strengths of the NEURO study included large population size, blinded imaging evaluations, required temporal proximity of late CUS and MRI. Limitations included that CUS was obtained by local centre protocol with possibility for technical inconsistencies, and there were low rates of adverse outcomes, leading to wide confidence intervals. While multivariable modelling taking into account neonatal risk factors scrutinises results, a drawback is potential over-analysis. Near-term CUS and MRI have been shown to be more reliable alternatives to early CUS in predicting outcomes, but imaging is only mildly effective in formulating predictions of neurodevelopmental outcomes in extremely premature neonates 2, 4, 9. There is still considerable uncertainty regardless of imaging modality or combination of imaging modalities used. Studies have shown that MRI before discharge is too early to detect posterior limb internal capsule myelination abnormalities, which have been shown to be associated with gross motor deficits 3, 5. Extending outcomes to include cognitive and behavioural difficulties of school-age ex-premature neonates will detect deficits not fully appreciated on Bayley or GMFCS testing 2, 9. https://ebneo.org/2018/06/late-neuroimaging-predicts-neurodevelopmental-outcomes-in-preterm-infants/ None. None.