One of the basic principles underlying modern neuroscience is that of connectional specificity; that is, neurons in different regions of the brain do not form connections randomly but rather in a manner that facilitates the processing of information among regions with related functions. A full characterization of the connectional specificity of the human brain would require identifying all connections between individual neurons and mapping the trajectories of the axons that connect them. This neuronal connectivity diagram, referred to as the human connectome, is unknown and beyond the reach of current technologies. However, recent advances in diffusion and functional MRI have opened the way to the study of major pathways in the brain, which are formed by bundles of axons running in parallel and terminating in groups of neurons with architectonic and functional homogeneity. These advances have brought us excitingly close to the in vivo, noninvasive mapping of the human connectome at the aggregate level but several technical challenges remain to be addressed.