Traditional paleodemographic methods of estimating mortality have been based on unrealistic assumptions about the prevalence of closed and stationary populations. When a living closed population was growing, the mean age at death of its skeletal survivors will be shifted below its true life expectancy. For declining populations, the mean age at death will be higher than true underlying life expectancy at birth. The faster the rate of growth, the larger and more curvilinear is the displacement. Mortality estimates can only be extracted from skeletal populations via an independent estimate of the growth rate. Fertility levels, however, can be estimated directly. The empirical importance of growth rate-dependent mortality estimates is demonstrated by reinterpreting mean-age-at-death data from several populations before and after the agricultural revolution; with detailed consideration given to the Old World populations of Acsadi and Nemeskeri and a New World population from central Illinois.