Hemodynamic signals, changes in cerebral blood flow, volume and oxygenation, are widely used to infer neural activity non-invasively, but their coupling of these signals to neural activity, and how this coupling might vary with behavioral state is still poorly understood. A quantitative understanding of the relationship between neural activity and hemodynamic signals across behavioral conditions is critical for the correct interpretation of neuroimaging studies. I will describe recent work from my lab investigating the relationship between cerebral hemodynamic signals and neural activity across behavioral states using optical imaging and electrophysiological approaches in awake, behaving mice. Surprising, we have found that hemodynamic signals are not always and everywhere faithful reporters of neural activity. These results have important implications for how we view the interactions between neurons and the vascular infrastructure that supplies them.