Tau-bearing neurofibrillary lesions present a promising biomarker for premortem diagnosis and staging of Alzheimer's disease and certain forms of frontotemporal lobar degeneration by whole brain imaging methods. Although brain penetrating compounds capable of binding tau aggregates with high affinity have been disclosed for this purpose, the major barrier to progress remains the need for tau lesion binding selectivity relative to amyloid-beta plaques and other deposits of proteins in cross-beta-sheet conformation. Here we discuss challenges faced in the development of tau lesion-selective imaging agents, and recent preclinical advances in pursuit of this goal.