Aims/hypothesis The primary aim of this study was to determine whether the presence of one or more APOE ε4 alleles modifies the association between diabetes (defined by glucose ≥7 mmol/l or treatment) and cognitive function. Methods Diabetic status and APOE genotype interactions were assessed cross-sectionally for 826 communitydwelling, stroke-free, non-demented individuals (526 nondiabetic non-APOE ε4 carriers, 174 non-diabetic APOE ε4 carriers, 87 diabetic APOE ε4 non-carriers, 39 diabetic APOE ε4 carriers) ranging in age from 50 to 98 years. Cognitive function was assessed using the Mini-Mental State Examination (MMSE), the similarities subtest from the Wechsler Adult Intelligence Scale, and four composite scores derived from 17 additional neuropsychological tests. Multiple linear regression analyses were employed to relate diabetes and APOE genotype to cognitive performance and to examine the interaction between these two risk factors as they relate to cognitive performance. Multiple cardiovascular disease risk factors were statistically controlled. Results With adjustment for age, education, sex, race/ ethnicity and APOE genotype, performance level was lower for the diabetic than for the non-diabetic group for the MMSE, the similarities subtest and each of the cognitive composites with the exception of the verbal memory composite. Interactions (p<0.05) between diabetes and APOE genotype were found for all but the visual-spatial memory/organisation composite. The negative association between diabetes and cognitive performance was of a higher magnitude for individuals who carry one or more APOE ε4 alleles. Results were similar with additional adjustment for cardiovascular disease and associated risk factors. Conclusions/interpretation The presence of one or more APOE ε4 alleles modifies the association between diabetes and cognitive function.