In this study, we examined the association of fasting serum insulin (FSI) and fasting serum glucose (FSG) with cognitive impairment in the very elderly using a sample of Chinese nonagenarians/centenarians. This study used data from a survey that was conducted in 2005 on all residents aged 90 years or more in a district with 2,311,709 inhabitants. FSG, FSI, and cognitive function were analyzed. The sample included 661 unrelated Chinese individuals (aged 90-108 years; mean, 93.52 ± 3.37 years; 67.17 % women; FSI, 6.27 ± 2.27 mU/mL; FSG levels, 4.46 ± 1.45 mmol/L). The prevalence of cognitive impairment was 61.81 % and that of hypoinsulinemia was 31.92 %. Individuals with hypoinsulinemia showed lower cognitive function scores (14.81 ± 5.79 vs. 15.78 ± 5.24, t = 2.160, P = 0.031). No differences in cognitive function score between different FSI and FSG groups were significant, and no differences in FSI and FSG between individuals with and without cognitive impairment were statistically significant. Unadjusted multiple logistic regressions showed that hypoinsulinemia, impaired fasting glucose, or diabetes did not change the risk of cognitive impairment significantly. In summary, we found that in elderly subjects, cognitive function appeared associated with FSI, and higher FSI may be associated with enhanced cognitive function.