When a volcano becomes restless, a primary question is whether the unrest will lead to an eruption. Here we recognize four possible outcomes of a magmatic intrusion: "deep intrusion", "shallow intrusion", "sluggish/viscous magmatic eruption", and "rapid, often explosive magmatic eruption". We define "failed eruptions" as instances in which magma reaches but does not pass the "shallow intrusion" stage, i.e., when magma gets close to, but does not reach, the surface. Competing factors act to promote or hinder the eventual eruption of a magma intrusion. Fresh intrusion from depth, high magma gas content, rapid ascent rates that leave little time for enroute degassing, opening of pathways, and sudden decompression near the surface all act to promote eruption, whereas decreased magma supply from depth, slow ascent, significant enroute degassing and associated increases in viscosity, and impingement on structural barriers all act to hinder eruption. All of these factors interact in complex ways with variable results, but often cause magma to stall at some depth before reaching the surface. Although certain precursory phenomena, such as rapidly escalating seismic swarms or rates of degassing or deformation, are good indicators that an eruption is likely, such phenomena have also been observed in association with intrusions that have ultimately failed to erupt. A perpetual difficulty with quantifying the probability of eruption is a lack of data, particularly on instances of failed eruptions. This difficulty is being addressed in part through the WOVOdat database. Papers in this volume will be an additional resource for scientists grappling with the issue of whether or not an episode of unrest will lead to a magmatic eruption.