Functional MRI (fMRI) based on the blood oxygenation level dependent (BOLD) contrast often suffers from a lack of specificity because of the vascular spread of oxygenation changes. It is suggested from the optical imaging and animal fMRI literature that cerebral blood volume (CBV) changes are more closely tied to the smaller vessels. As such, fMRI contrast based on CBV changes will have improved spatial specificity to the neuronal activities as they are immediately adjacent to the smaller vessels. In this paper, an endogenous contrast mechanism based on a diffusion weighting strategy that could detect functional CBV changes is presented. Initially, a theoretical framework is presented to model the functional signal changes as a function of CBV under diffusion weighting, which predicts peak CBV sensitivity at various vessel–tissue mixtures. It was found that a b factor over 1500 s/mm2 would be necessary to achieve dominant CBV contrast. Further, two sets of experimental results are also presented. In the first experiment, diffusion weighting at a set of b factors ranging from 300 to 600 s/mm2 was used. The results indicated that while the positive activation (predominantly BOLD signal) continued to reduce in magnitude and spatial extent, the negative activation (predominantly CBV signal) remained virtually constant with increasing b factors. The second experiment used a b factor of 1600 s/mm2 and showed extensive negative activation in the visual cortex and greatly reduced positive activations compared with images with no diffusion weighting. The time course of negative activation showed a faster time to peak and return to baseline than the positive BOLD activity, consistent with the small vessel origin of the signal changes. These results suggest that appropriate diffusion weighting could be used to measure activation related CBV changes. Copyright © 2006 John Wiley & Sons, Ltd.