Caffeine ingestion results in increased brain cell metabolism (Nehlig et al.  Brain Res Brain Res Rev 17:139–170) and decreased cerebral blood flow (Field et al.  Radiology 227:129–135; Mulderink et al.  Neuroimage 15:37–44). The current study investigated the effect of caffeine in a word stem completion task using only novel word stems (no repeated stimuli). Resting perfusion was measured with arterial spin labeled perfusion MRI, along with blood oxygenation level-dependent (BOLD) signal before and after ingestion of regular coffee, decaffeinated coffee, and water. Based on previous research (Laurienti et al.  Neuroimage 17:751–757; Mulderink et al.  Neuroimage 15:37–44), we hypothesized that caffeine would result in increased BOLD signal intensity and extent of BOLD activation. As expected, caffeine resulted in a significant decrease in cerebral perfusion. However, both the control and caffeine groups showed an increase in BOLD signal amplitude across two sets of novel word stems. Additionally, the control group showed a 50% reduction in the extent of BOLD activation, while the caffeine group showed no change in activation extent. Neither group showed changes in BOLD baseline signal over time, which had been suggested to mediate caffeine-related BOLD signal changes. The results suggest that caffeine may attenuate general task practice effects that have been described in recent functional MRI studies of word stem completion (Buckner et al.  Brain 123:620–640). Hum Brain Mapp 2006. © 2006 Wiley-Liss, Inc.