The authors examined national changes in socioeconomic differentials in mortality for middle-aged and older white men and women in the United States with the use of 1960 data from the Matched Records Study and 1971-1984 data from the first National Health and Nutrition Examination Survey (NHANES I) Epidemiologic Follow-up Study (NHEFS). In 1960, there was little difference in mortality by educational level among middle-aged and older men. Since 1960, death rates among men declined more rapidly for the more educated than the less educated, which resulted in substantial educational differentials in mortality in 1971-1984. In contrast, among women, death rates declined at about the same rate regardless of educational attainment, so that a strong inverse relation between education and mortality in 1960 remained about the same magnitude during 1971-1984. Trends in educational differentials for heart disease mortality are responsible for much of the change for all causes of death. Relative risk estimates based on the NHEFS indicate that after taking into account selected baseline risk factors the least educated are still at substantially elevated risk of death from heart disease, ranging from a relative risk of 1.38 for men aged 65-74 years at baseline to 2.27 for men aged 45-64 years. Reasons for the observed educational differentials and their changes over time are not easily explained and are likely to be multifactorial.