OBJECTIVE: To examine the temporal relation between depression and cognitive impairment in old age. DESIGN: Prospective, population based study with four years of follow up. SETTING: City of Leiden, the Netherlands. PARTICIPANTS: 500 people aged 85 years at recruitment. MAIN OUTCOME MEASURES: Annual assessments of depressive symptoms (15 item geriatric depression scale), global cognitive function (mini-mental state examination), attention (Stroop test), processing speed (letter digit coding test), and immediate and delayed recall (12 word learning test). RESULTS: At 85 years old, participants' depressive symptoms and cognitive impairment were highly significantly correlated (P < 0.001). During follow up, an accelerated annual increase of depressive symptoms was associated with impaired attention (0.08 points (95% confidence interval 0.01 to 0.16)), immediate recall (0.17 points (0.09 to 0.25)), and delayed recall (0.10 points (0.02 to 0.18)) at baseline. In contrast, depressive symptoms at baseline were not related to an accelerated cognitive decline during follow up (P > 0.05). CONCLUSION: Caregivers should be aware of the development of depressive symptoms when cognitive impairment is present. However, the presence of depression only does not increase the risk of cognitive decline.