Objectives: Although previous studies have extensively confirmed the cross-sectional relationship between cognitive impairment and depression in depressed elderly patients, the findings of their longitudinal associations are still mixed. The purpose of this study was to explore the two-way causal relationship between depression symptoms and cognition in patients with late-life depression (LLD).Methods: A total of 90 patients with LLD were assessed across two time points (baseline and 1-year follow up) on measures of 3 aspects of cognition and depressive symptoms. The data were then fitted to a structural equation model to examine two cross-lagged effects.Results: Depressive symptoms predicted a decline in executive function (β = 0.864, p = 0.049) but not vice versa. Moreover, depressive symptoms were predicted by a decline in scores of working memory test (β = −0.406, p = 0.023), respectively. None of the relationships between the two factors was bidirectional.Conclusion: These results provide robust evidence that the relationship between cognition and depressive symptoms is unidirectional. Depressive symptoms may be a risk factor for cognitive decline. The decrease of information processing speed predicts depressive symptoms.