BACKGROUND: Growing evidence suggests that cerebral white-matter changes and depressive symptoms are linked directly along the causal pathway. We investigated whether baseline severity of cerebral white-matter changes predict longer-term future depressive outcomes in a community sample of non-disabled older adults. METHOD: In the Leukoaraiosis and Disability in the Elderly (LADIS) study, a longitudinal multi-centre pan-European study, 639 older subjects underwent baseline structural magnetic resonance imaging (MRI) and clinical assessments. Baseline severity of white-matter changes was quantified volumetrically. Depressive outcomes were assessed in terms of depressive episodes and depressive symptoms, as measured by the Geriatric Depression Scale (GDS). Subjects were clinically reassessed annually for up to 3 years. Regression models were constructed to determine whether baseline severity of white-matter changes predicted future depressive outcomes, after controlling for confounding factors. RESULTS: Baseline severity of white-matter changes independently predicted depressive symptoms at both 2 (p<0.001) and 3 years (p=0.015). Similarly, white-matter changes predicted incident depression (p=0.02). Over the study period the population became significantly more disabled (p<0.001). When regression models were adjusted to account for the influence of the prospective variable transition to disability, baseline severity of white-matter changes no longer predicted depressive symptoms at 3 years (p=0.09) or incident depression (p=0.08). CONCLUSIONS: Our results support the vascular depression hypothesis and strongly implicate white-matter changes in the pathogenesis of late-life depression. Furthermore, the findings indicate that, over time, part of the relationship between white-matter changes and depression may be mediated by loss of functional activity.