Empirical “value-of-life” estimates derived from labor market wage-risk premiums have varied widely. This paper examines the influence of risk variable definition on these estimates. Value-of-life estimates are derived for one sample from a set of several risk measures. The analysis reveals that the risk variable definition can markedly affect the value-of-life estimate. Further, the paradoxical pattern of “high” estimates from industry risk data and “low” estimates from occupation data is shown to be attributable to different risk definition, not differences in the characterization of risk faced by a given sample of works. Finally, by associating consistently low estimates with one particular risk variable this study suggests much more uniformity in the value-of-life estimates than previously believed.