The aim of this study is to analyze how relative mortality risk varies between persons with and without heart problems in different social classes. Cox proportional hazards regression was used to analyze relative mortality risk for the period 1968-1996 for a Swedish nationally representative sample of 4585 persons born between 1892 and 1942, and interviewed 1968. Survivors from the original sample were also interviewed in 1974, 1981 and 1991 or 1992. "Heart problems" is defined as the presence of three mild or one severe symptom associated with circulatory problems. Social class is based on occupation. The relative mortality risk varied significantly between social classes and between persons with and without heart problems, among both men and women. These differences were smaller among women than among men. The main results are that there are significant additive interactions between social class and heart problems among men. Men from lower social classes have a more elevated mortality risk than men from higher social classes when they have a heart problem. Among white-collar workers the coefficient of the difference between men with and without heart problems was 0.53. The corresponding difference was significantly larger among workers (1.59, P = 0.01), thus demonstrating an additive interaction. The difference was even greater (1.86) among "unclassifiable" men--those who could not report an occupation that could be coded into a social class, mainly because they were long-term unemployed or on early-retirement pensions. Among women, the mortality difference between white-collar workers with and without heart problems was 0.85. None of the mortality differences between those with and without heart problems in other social classes differed significantly from those of white-collar workers. The mortality difference between women with and without heart problems was, however, large (2.34) among the "unclassifiable". This difference is even larger than the corresponding difference among men.