Since it was introduced a decade ago, functional magnetic resonance imaging (fMRI) has come to dominate research on the human brain. However, fMRI maps are based on secondary metabolic and hemodynamic events that follow neuronal activity, and not on the electrical activity itself. Therefore, the representation provided by fMRI cannot be assumed a priori to be exact. The accuracy of these maps depends on the spatial extent of the metabolic and hemodynamic changes induced by neuronal activity, and the role played by the vasculature in converting these changes to signals detected by magnetic resonance imaging. Significant progress has been made in both areas, suggesting that it is possible to obtain both spatially accurate and quantitative data on brain function from magnetic resonance methodologies.