A cohort of patients with first-episode schizophrenia was dichotomised into two age- and sex-matched groups of clinical syndromes, the active and withdrawn, and underwent high-resolution three-dimensional magnetic resonance imaging at baseline and 8 months later. A cohort of age- and sex-matched normal controls was also imaged at the same time intervals. The application of subvoxel registration and semiautomated quantification techniques demonstrated a significantly different outcome in ventricular changes between the two groups of patients. Compared with the controls, the withdrawn patients showed progressive ventricular enlargement, with an increase in ventricle-to-brain volume ratio, whereas the active group showed a reduction in ventricle-to-brain volume ratio, with a change opposite in sign and smaller in magnitude. These findings lend further support for the aetiological validity of this syndromal model of schizophrenia and are likely to be of importance in furthering our understanding of its pathogenesis and in the development of suitable therapeutic strategies.