OBJECTIVES: Cognitive impairment is an important consequence of sepsis. We sought to determine long-term trajectories of cognitive function after sepsis. DESIGN: Prospective study of the Reasons for Geographic and Racial Differences in Stroke cohort. SETTING: United States. PATIENTS: Twenty-one thousand eight-hundred twenty-three participants greater than or equal to 45 years, mean (sd) age 64.3 (9.2) years at first cognitive assessment, 30.9% men, and 27.1% Black. MEASUREMENTS AND MAIN RESULTS: The main exposure was time-dependent sepsis hospitalization. The primary outcome was global cognitive function (Six-Item Screener range, 0-6). Secondary outcomes were incident cognitive impairment (Six-Item Screener score ≤ 4 [impaired] vs ≥5 [unimpaired]), new learning (Consortium to Establish a Registry for Alzheimer Disease Word List Learning range, 0-30), verbal memory (word list delayed recall range, 0-10), and executive function/semantic fluency (animal fluency test range, ≥ 30). Over a median follow-up of 10 years (interquartile range, 6-12 yr), 840 (3.8%) experienced sepsis (incidence 282 per 1,000 person-years). Sepsis was associated with faster long-term declines in Six-Item Screener (-0.02 points per year faster [95% CI, -0.01 to -0.03]; p < 0.001) and faster long-term rates of incident cognitive impairment (odds ratio 1.08 per year [95% CI, 1.02-1.15]; p = 0.008) compared with presepsis slopes. Although cognitive function acutely changed after sepsis (0.05 points [95% CI, 0.01-0.09]; p = 0.01), the odds of acute cognitive impairment (Six-Item Screener ≤ 4) immediately after sepsis was not significant (odds ratio, 0.81 [95% CI, 0.63-1.06]; p = 0.12). Sepsis hospitalization was not associated with acute changes or faster declines in word list learning, word list delayed recall, or animal fluency test. CONCLUSIONS: Sepsis is associated with accelerated long-term decline in global cognitive function.