Aims: an association between T2DM and cognitive impairment has been reported among Western populations, but data are limited in other settings. We investigated the cross-sectional association between fasting blood glucose (FBG) and cognition in an older Chinese population. Methods: the Guangzhou Biobank Cohort Study included 27,971 individuals (50–96 years, mean age 61.5 years, 72% female) with measures of cognitive function assessed using modified Delayed Word Recall Test (DWRT) and Mini-Mental State Examination (MMSE). Fasting glucose and lipids, and potential confounders were measured. Results: after adjustment for potential confounders, the risk for cognitive impairment as measured by DWRT, significantly increased [odds ratio (OR) = 1.18, 95% CI 1.00–1.40] but the association was of borderline significance when measured by MMSE (OR = 1.04, 95% CI 0.73–1.47) in those with diabetes relative to those without diabetes. Fasting blood glucose was significantly negatively associated with cognitive function as measured by DWRT but not MMSE, with an increase of 1 mmol/l of FBG associated with a decrease of 0.02 in DWRT (P < 0.05, 95% CI −0.03 to −0.002) and 0.03 in MMSE score (P = 0.114, 95% CI −0.06–0.01). Conclusions: an FBG level indicative of T2DM was associated with increased risk for cognitive impairment. The findings also demonstrate that glycaemia is continuously associated with cognitive impairment, suggesting that dysfunction is associated with increasing glucose levels even in the normoglycaemic range.