Diffusion tensor imaging (DTI) utilizes the diffusion of water to create detailed maps of brain white matter structure. DTI provides information not apparent on conventional magnetic resonance imaging, and has proved to be more sensitive to disruption of white matter development than any other imaging method. In children with cerebral palsy, DTI can add anatomical specificity and quantitative information about the white matter tract size and microstructure. It has been of great benefit in children born preterm with periventricular leukomalacia (PVL).1 Studies have variably shown correlations between corticospinal tract injury2 and thalamocortical projections (posterior thalamic radiation)3,4 in the pathophysiology of motor disability in these children. Yoshida et al.5 reported that when compared with controls, children with PVL have reduced fractional anisotropy and number of fibers in the region of corticospinal tract injury and posterior thalamic radiation, and that these measurements correlate with a clinical measure of function. These findings are a quantitative confirmation of the qualitative data from previous studies and from the visual impression of experienced clinicians. This categorization of variable injury in sensory and motor pathways provides further support for the development of rehabilitative interventions that selectively target affected pathways. The interpretation of results from DTI studies is predicated on an understanding of the strengths and limitations of the methodology employed. The study by Yoshida et al. illustrates important questions regarding the interpretation of quantitative DTI data and the limitations of analysing individual tracts in disorders affecting multiple brain regions such as in children with PVL.6 In regard to interpretation of DTI data, three points need to be carefully considered. First, brain volume and diffusivity have a strong time-dependence, particularly in the first 2 years of life; therefore, studies involving participants in their infancy and childhood may be influenced by this age effect. Second, measurements based on regions of interest are largely evaluator-dependent and interrater agreement should always be presented, as it is an important indicator of the reliability and reproducibility of the results. Finally, measures of imaging contrast and volume are interrelated, as the fractional anisotropy threshold setting will partially determine the volume calculated and when regions of interest are generated, the volume will influence the fractional anisotropy measurements. This type of coupling of the ‘anatomical definition’ and ‘intensity shift’ is a common issue in image analysis and requires careful interpretation. DTI tractography will only describe specific tracts. As the injury in children born preterm is varied and diffuse to both gray and white matter, a global and integrative analysis of neuroanatomy is essential for understanding the pathophysiology of PVL. Recently, state-of-the-art non-linear normalization techniques, such as large-deformation diffeomorphic metric mapping, have been developed to obtain global and automated analysis of the whole brain.7 When combined with other imaging methods, such as functional resting connectivity,8 these tools have the potential to describe the pathophysiology of brain injury in children with cerebral palsy and other childhood neurological disorders accurately and automatically. This is an essential step for detailed anatomy function correlation studies and to provide crucial information that can be useful in intervention studies and for the determination of prognosis.