Visual attention can enhance visual performance. Exactly how this is accomplished remains unclear. Although previous fMRI studies have shown that attention can modulate activity throughout visual cortex, they have not distinguished between two possibilities of this modulation: attention could increase the input contrast (contrast gain) or amplify the output signal (response amplification). Psychophysical and neurophysiological experiments have attempted to distinguish between these possibilities. However, psychophysical experiments can only infer the effect of attention on the visual system as a whole, and physiology can only look at properties of a small number of neurons. To distinguish between contrast gain and response amplification models, it is necessary to measure the effect of attention on a range of input contrasts. Assuming that BOLD signal is proportional with neural activity, a contrast gain control model would predict that attention will shift the contrast response function horizontally on the log contrast axis. Alternatively, the output response amplification model would predict that attention will shift the contrast response function vertically on the log BOLD axis. Using event-related design, subjects viewed two identical gratings of various contrasts (from 3% to 99%) presented to the left and right of fixation for 2 sec every 20 sec. In each trial, before grating onset, a small arrow was presented briefly at the fixation point to direct subject's attention. Subjects were instructed to shift their attention to the cued grating while maintaining fixation. Retinotopic mapping was used to identify individual visual areas, and BOLD contrast response functions in these areas were compared between the attended and non-attended stimulus. Results from three subjects show that attention enhanced BOLD responses to both low and high contrast gratings. These results suggest that attention may increase the contrast sensitivity AND also amplifies the response.