This article presents an analysis of trends in mortality differentials and life expectancy by average relative earnings for male Social Security-covered workers aged 60 or older. Because average relative earnings are measured at the peak of the earnings distribution (ages 45-55), it is assumed that they act as a rough proxy for socioeconomic status. The historical literature reviewed in this analysis generally indicates that mortality differentials by socioeconomic status have not been constant over time. For this study, time trends are examined by observing how mortality differentials by average relative earnings have been changing over 29 years of successive birth cohorts that encompass roughly the first third of the 20th century. Deaths for these birth cohorts are observed at ages 60-89 from 1972 through 2001, encompassing roughly the last third of the 20th century. The large size and long span of death observations allow for disaggregation by age and year-of-birth groups in the estimation of mortality differentials by socioeconomic status. This study finds a difference in both the level and the rate of change in mortality improvement over time by socioeconomic status for male Social Security-covered workers. Average relative earnings (measured as the relative average positive earnings of an individual between ages 45 and 55) are used as a proxy for adult socioeconomic status. In general, for birth cohorts spanning the years 1912-1941 (or deaths spanning the years 1972-2001 at ages 60-89), the top half of the average relative earnings distribution has experienced faster mortality improvement than has the bottom half. Specifically, male Social Security-covered workers born in 1941 who had average relative earnings in the top half of the earnings distribution and who lived to age 60 would be expected to live 5.8 more years than their counterparts in the bottom half. In contrast, among male Social Security-covered workers born in 1912 who survived to age 60, those in the top half of the earnings distribution would be expected to live only 1.2 years more than those in the bottom half. The life expectancy estimates in this article represent one possible outcome under one set of assumptions. These projections should not be regarded as an accurate depiction of the future. Specifically, this study adopts a simple projection method in which differentials are assumed to follow the pattern observed over the last 30 years of the 20th century for the first 30 years of the 21st century. This assumption lacks theoretical underpinnings because the causes of the widening differentials observed over the past 30 years have not been determined. On the one hand, if the trend of widening mortality differentials by year of birth observed over the past 30 years does not continue, the projection method used in this analysis could lead to an overestimation of future differences in life expectancy between socioeconomic groups. On the other hand, if mortality differentials do not narrow by age as observed in the past, the projection method used could lead to an underestimation of the differences in life expectancy between socioeconomic groups aged 60 or older.