This repeated-measures study sought to determine whether hypercapnic normalization can be used to address caffeine-induced vascular changes affecting BOLD responses. Visual/motor BOLD activation and cerebrovascular reactivity (CVR) under caffeine and placebo conditions were examined. CVR-correction was performed by dividing the task BOLD signal by CVR. Prior to CVR-correction, peak BOLD responses for both caffeine and placebo conditions were found in the visual cortex. After CVR-correction, peak responses were in the subcortical regions. Caffeine produced larger responses in the cerebellum compared to placebo, while subcortical activation was larger under placebo conditions. These results show that CVR-correction may benefit subcortical signal detection.