Caffeine is the most widely consumed psychoactive substance, yet few studies have investigated how habitual and acute consumption and withdrawal impacts resting-state brain activity. Notably, prior research lacks adequate control for deprivation state, despite evidence that caffeine reinforcement occurs primarily by alleviating withdrawal. This study used a between-participants design to assess resting-state fMRI brain activity, mood, and cognition in three groups: (1) moderate consumers (200-500mg/day) tested after overnight abstinence (caffeine withdrawn, CW); (2) moderate consumers tested after overnight abstinence followed by 100mg of caffeine (caffeine not-withdrawn, CNW); and (3) non-consumers of caffeine (<50mg/day, NC). Sixty healthy volunteers, aged 18-45 (n = 20/group) completed the Bond-Lader mood battery, a rapid visual information processing task and a resting-state fMRI scan. For resting-state brain activity, the CW group showed altered nucleus accumbens connectivity with primary visual cortex compared to CNW and NC groups. The CNW group showed stronger anterior insula connectivity with precuneus cortex compared to CW and NC groups. For network-level analyses, the CNW group exhibited reduced limbic within-network connectivity and altered connectivity between limbic and occipital cortex compared to CW and NC groups. The anterior salience network showed group differences in connectivity with the putamen, pallidum, and thalamus. The supplementary somatomotor network showed greater connectivity with the bilateral putamen in both caffeine groups, but reduced connectivity with the right middle temporal gyrus for the CW group. No significant main group effect emerged for mood and cognition. These findings demonstrate that caffeine consumption and withdrawal produce distinct alterations in resting-state brain activity.