Cognitive dysfunction is common in older persons suffering from a major depression. However, the degree to which this dysfunction is reversible with successful treatment of the depression remains uncertain. The present study examined the effects that treatment (randomized double-blind design) with either an SSRI (paroxetine) or a tricyclic antidepressant (nortriptyline) had on cognition in older depressed patients. The patients' performance was compared to that of a group of normal controls of similar age and education. Patients and controls were administered measures of working memory, information-processing speed, episodic memory and attention five times over the course of a 12 week trial. At baseline, the patients performed more poorly than the elderly controls on all cognitive measures. While the patients' performance did improve over the course of their treatment, the magnitude of this improvement did not exceed that produced in the elderly controls by practice alone. The same pattern of results was evident in both intent-to-treat and responder analyses. Thus, there was no evidence that the depressed patients' cognitive performance normalized after response to antidepressant therapy. Neither the patients' age at onset nor their baseline level of cognitive functioning influenced the amount by which their performance improved over the 12 week trial. There was no difference between paroxetine and nortriptyline in the amount of cognitive change associated with treatment. The present results suggest that cognitive dysfunction persists in older depressed patients even after their mood disorder has responded to antidepressant medications.