Flavanol consumption is favorably associated with cognitive function. We tested the hypothesis that dietary flavanols might improve cognitive function in subjects with mild cognitive impairment. We conducted a double-blind, parallel arm study in 90 elderly individuals with mild cognitive impairment randomized to consume once daily for 8 weeks a drink containing ≈990 mg (high flavanols), ≈520 mg (intermediate flavanols), or ≈45 mg (low flavanols) of cocoa flavanols per day. Cognitive function was assessed by Mini Mental State Examination, Trail Making Test A and B, and verbal fluency test. At the end of the follow-up period, Mini Mental State Examination was similar in the 3 treatment groups (P=0.13). The time required to complete Trail Making Test A and Trail Making Test B was significantly (P<0.05) lower in subjects assigned to high flavanols (38.10±10.94 and 104.10±28.73 seconds, respectively) and intermediate flavanols (40.20±11.35 and 115.97±28.35 seconds, respectively) in comparison with those assigned to low flavanols (52.60±17.97 and 139.23±43.02 seconds, respectively). Similarly, verbal fluency test score was significantly (P<0.05) better in subjects assigned to high flavanols in comparison with those assigned to low flavanols (27.50±6.75 versus 22.30±8.09 words per 60 seconds). Insulin resistance, blood pressure, and lipid peroxidation also decreased among subjects in the high-flavanol and intermediate-flavanol groups. Changes of insulin resistance explained ≈40% of composite z score variability through the study period (partial r(2)=0.4013; P<0.0001). To the best of our knowledge, this is the first dietary intervention study demonstrating that the regular consumption of cocoa flavanols might be effective in improving cognitive function in elderly subjects with mild cognitive impairment. This effect appears mediated in part by an improvement in insulin sensitivity.