Previous studies have demonstrated that patients with major depression following stroke have a greater degree of cognitive impairment than nondepressed patients with comparable lesions. The present study examined the longitudinal course of cognitive impairment related to depression. Patients were prospectively evaluated following an acute stroke (N = 309), using a structured psychiatric interview and the Mini-Mental State Examination. Longitudinal evaluations were obtained at 3, 6, 12, and 24 months follow-up in a subset of these patients. During the initial in-hospital evaluation, the frequency and severity of cognitive impairment was significantly greater in patients with major depression compared with nondepressed patients. This effect occurred predominantly in patients with major depression following left hemisphere stroke. The association of depression and cognitive function was strongest during the initial evaluation, but was present for up to 1 year. The year-long effect, however, was evident only in patients with left hemisphere stroke. Patients with both depression and cognitive impairment had a greater duration of depression than depressed patients without cognitive impairment. Depression with cognitive impairment appears to be a phenomenon produced by left hemisphere lesions. This suggests that left hemisphere stroke may produce depression through a different mechanism than lesions in other locations. In addition, the fact that the strongest influence of depression on cognitive function was seen during the initial evaluation suggests that this phenomenon may be mediated by acute or subacute physiological effects of the lesion.