Neurodegenerative diseases are characterized by 'hot spots' of degeneration. The regions of primary vulnerability vary between different neurodegenerative diseases. Within these regions, some neurons are lost whereas others that are morphologically indiscriminable survive. The enigma of this selective vulnerability is tightly linked to two fundamental problems in the neurosciences. Firstly, it is not understood how many neuronal cell types make up the mammalian brain; estimates are in the order of more than thousand. Secondly, the mechanisms by which some nerve cells undergo functional impairment followed by degeneration while others do not, remain elusive. Understanding the basis for this selective vulnerability has significant implications for understanding the pathogenesis of disease and for developing treatments. Here, we review what is known about selective vulnerability in Alzheimer's disease, frontotemporal dementia and Parkinson's disease. We suggest, since transgenic animal models of disease reproduce aspects of selective vulnerability, that these models offer a valuable system for future investigations into the physiological basis of selective vulnerability.