BACKGROUND: The aim of this paper is to show for the first time mortality differentials by level of education for Swiss men and women. This work is of interest to public health efforts in Switzerland as well as for co-operative international research into the determinants of socioeconomic differentials in health and mortality. METHODS: This study is based on a longitudinal data set from the Swiss National Cohort, currently incorporating a probabilistic record linkage of the 1990 Swiss census, and all subsequent deaths until the end of 1997. The study population covers all Swiss nationals aged >/=25 years living in German speaking Switzerland, with 19.7 million person-years and 296 929 deaths observed. Educational gradients were analysed using standardized mortality ratios, multiple logistic regression, and the Relative Index of Inequality (RII). RESULTS: There were sizeable gradients in mortality by education for all age groups and both sexes. The mortality odds ratio decreased by 7.2% (95% CI: 7.0-7.5%) per additional year of education for men, and by 6.0% (95% CI: 5.6-6.3%) for women. In men, we found a steady decrease of the gradient from 13.1% (95% CI: 11.9-14.4%) in the age group 25-39 to 4.5% (95% CI: 4.0-5.0%) in the age group >/=75 years. For women in the age groups under 65 the gradients were smaller; over the age of 40 there was no decrease with increasing age. These results were fairly insensitive to variations in the parameters of record linkage. CONCLUSIONS: Despite a comparatively low overall mortality, Swiss men in the 1990s show larger relative gradients in mortality by education than men in other European countries in the 1980s, with the possible exception of younger men in Italy. In Switzerland there is a sizeable potential for further increasing overall life expectancy by reducing the mortality of those with a lower educational level. The results presented contribute to a reliable assessment of socioeconomic mortality differentials in Europe.