Objective: Cognitive impairments are present immediately following recovery from a first episode of mania, although at a lesser severity than those seen in more chronic patients with bipolar I disorder. Little is known about how deficits evolve over the course of illness, however, and whether these changes are associated with disease progression. Method: Patients with bipolar I disorder (DSM-IVTR) receiving naturalistic clinical follow-up from the Systematic Treatment Optimization Program for Early Mania (STOP-EM) from July 2004 to May 2013 completed a comprehensive cognitive battery following recovery from their first manic episode and again 1 year later. Performance was compared between patients who experienced a recurrence of a mood episode (BD recur) (n = 26) versus those that maintained remission (BD well) (n = 27) over followup, as well as healthy comparison subjects (HS) (n = 31). Results: While both BDrecur and BDwell had impairments in overall cognitive performance relative to HS at baseline (mean difference = -0.59, P < .001; mean difference = -0.43, P < .05, respectively), at follow-up BDrecur showed deficits compared to both HS (mean difference = -0.62, P = .001) and BDwell (mean difference = -0.41, P = .05), with BDwell cognition similar to that in HS (mean difference = -0.21, P > .4). BDwell showed larger improvements over follow-up relative to both other groups (P < .05). While changes in BDrecur did not differ from HS, in this group more days in a manic or hypomanic episode was associated with performance declines (r = -0.40, P < .05). Conclusions: While cognitive function improves in patients who sustain remission in the year following a first manic episode, those who experience a recurrence remain impaired, with performance declines being most apparent in those who experienced longer manic or hypomanic episodes.