Introduction: Diabetes is prospectively associated with cognitive decline. Whether lower cognitive function and worse brain structure are prospectively associated with incident diabetes is unclear. Methods: We analyzed data for 10,133 individuals with cognitive function testing (1990–1992) and 1212 individuals with brain magnetic resonance imaging (1993–1994) from the Atherosclerosis Risk in Communities cohort. We estimated hazard ratios for incident diabetes through 2014 after adjustment for traditional diabetes risk factors and cohort attrition. Results: Higher level of baseline cognitive function was associated with lower risk for diabetes (per 1 standard deviation, hazard ratio = 0.94; 95% confidence interval = 0.90, 0.98). This association did not persist after accounting for baseline glucose level, case ascertainment methods, and cohort attrition. No association was observed between any brain magnetic resonance imaging measure and incident diabetes. Discussion: This is one of the first studies to prospectively evaluate the association between both cognitive function and brain structure and the incidence of diabetes.