Motivated by prior data on local cortical shrinkage and intracortical myelination, we predicted age-related changes in topological organisation of cortical structural networks during adolescence. We estimated structural correlation from magnetic resonance imaging measures of cortical thickness at 308 regions in a sample of N=297 healthy participants, aged 14-24 years. We used a novel sliding-window analysis to measure age-related changes in network attributes globally, locally and in the context of several community partitions of the network. We found that the strength of structural correlation generally decreased as a function of age. Association cortical regions demonstrated a sharp decrease in nodal degree (hubness) from 14 years, reaching a minimum at approximately 19 years, and then levelling off or even slightly increasing until 24 years. Greater and more prolonged age-related changes in degree of cortical regions within the brain network were associated with faster rates of adolescent cortical myelination and shrinkage. The brain regions that demonstrated the greatest age-related changes were concentrated within prefrontal modules. We conclude that human adolescence is associated with biologically plausible changes in structural imaging markers of brain network organization, consistent with the concept of tuning or consolidating anatomical connectivity between frontal cortex and the rest of the connectome.Human adolescence is known to be a major phase of cortical development. In particular, cerebral cortex becomes thinner (Wierenga et al., 2014) and more densely myelinated in the transition from puberty to young adulthood. Adolescent decreases in cortical thickness (thinning) are variable between different areas of cortex (Raznahan et al., 2011): for example, thinning is greater in association cortical areas than primary sensory areas (Whitaker, Vértes et al., 2016).Motivated by these and other results, we predicted that human adolescence should be associated with changes in the architecture of structural brain networks. There are currently only two experimental techniques, both based on magnetic resonance imaging (MRI), that are capable of providing data to test this prediction: diffusion tensor imaging followed by tractography; or structural MRI followed by structural covariance or correlation analysis. Here we focused on the latter, measuring the thickness of a set of predefined cortical regions in each individual MRI dataset and then estimating the correlation of thickness between each possible pair of regions across participants. Similar methods have been widely used and validated (Lerch et al., 2006) in a range of prior studies Evans, 2013).In particular, structural correlation (covariance) measures have been used as a basis for graph theoretical modelling of the human connectome (Bullmore & Sporns, 2009;Fornito et al., 2016). Considerable evidence has accumulated in support of the general view that human brain structural correlation networks have a complex topological organization, characterised by...