We performed a randomized, double blind, placebo-controlled trial with blueberry supplementation in older adults with mild cognitive impairment. We measured cognitive performance and parent (unmodified food form) anthocyanins and a suite of flavonoid-based phase 2 metabolites arising from anthocyanins in urine. The blueberry-treated group exhibited improved semantic access (p = 0.01) and visual-spatial memory (p = 0.05), and there was a trend for enhanced speed of processing (p = 0.08). There was no group difference in urinary excretion of total anthocyanins (parent plus metabolite forms) due to an abundance of phase 2 metabolites in both groups. However, parent anthocyanins (less than 0.1% of total) were 100 times greater in the blueberry group and were correlated with neurocognitive benefit. Ongoing blueberry intake introduced in advance of dementia improved cognitive performance, which was correlated specifically with the recent intake of blueberry and an abundance of parent anthocyanins.