Lower socioeconomic groups experience higher rates of premature mortality in comparison with the upper socioeconomic groups. We have undertaken a study in Greece to assess socioeconomic differentials in overall, cancer, and cardiovascular (CVD) mortality and to identify their possible roots, using educational attainment to indicate socioeconomic status. Among participants in the general population, the Greek European Prospective Investigation into Cancer and Nutrition cohort, 23 697 individuals with no prevalent cancer or CVD disease at enrollment and with complete information on education and established or likely mortality risk factors, were followed up for an average of 9.6 years. Age-adjusted odds ratios of the prevalence of risk factors by education and sex were calculated through logistic regression and mortality ratios were estimated through Cox regression. With respect to overall and CVD mortality, the results indicated a 50% to more than 100% difference between the extreme categories of educational attainment. No gradient was, however, observed in cancer mortality. Obesity, poor diet, hypertension, and low physical activity were more prevalent among the least educated participants, but smoking was almost as common among the more and the less educated men and strikingly more common among the higher educated women. Likely intermediates did not explain more than one-third of the excess mortality among the less educated persons. In the Greek population, strong socioeconomic gradients were observed in the overall and CVD mortality, but not in cancer mortality. Established risk factors for premature mortality explained only a fraction of the observed gradients.