We agree that the process of establishing a definitive extent of white matter disruption in schizophrenia, and its relationship with illness duration and antipsychotic medication, is likely to be a lengthy one – larger studies such as ours notwithstanding. But we would like to clarify our reasoning with respect to Chua & McAlonan's comments and the methodological alternatives they suggest.

First, we suspect that a ‘core pathology’ for white matter abnormalities in schizophrenia may be rather more elusive than the reconciliation we attempted. Although the recent meta-analysis by Ellison-Wright & Bullmore ^{Reference Ellison-Wright and Bullmore1} has found areas of most common difference in the 15 studies they examined, it should be noted that only a fraction of the studies they looked at shared these differences – and the history of diffusion tensor imaging in schizophrenia is full of such conflicts. ^{Reference Kanaan, Kim, Kaufmann, Pearlson, Barker and McGuire2} Although there may indeed be areas of greater difference, the evidence is against any difference that is common to all.

Second, with regard to distinguishing the effects of duration of illness and antipsychotic medication, drug-naive cohorts clearly offer enormous potential as the authors acknowledge. Such cohorts are difficult to obtain in high-income countries however, and the alternative approach they suggest – of ANOVA with covariation – has similar difficulties, since duration of treatment and illness will be so strongly correlated in most samples. We also note that the studies they cite as demonstrating the effectiveness of this approach either did not covary for medication exposure ^{Reference Szeszko, Ardekani, Ashtari, Kumra, Robinson and Sevy3} or did not use diffusion tensor imaging. ^{Reference Chua, Deng, Chen, Law, Chiu and Cheung4}