Over the past 40 years Estonia has experienced similar developments in mortality to other former Soviet countries. The stagnation in overall mortality has been caused mainly by increasing adult mortality. However, less is known about the social variation in health. This study examines differences in self-rated health by eight main dimensions of the social structure on the basis of the Estonian Health Interview Survey, carried out in 1996/1997. A multistage random sample (n = 4711) of the Estonian population aged 15-79 was interviewed; the response rate was 78.3%. This study includes those respondents aged 25-79 (n = 4011) with analyses being performed separately for men and women. The study revealed that a low educational level, Russian nationality, low personal income and for men only, rural residence were the most influential factors underlying poor health. Education had the biggest independent effect on health ratings: for women with less than an upper secondary education the odds of having poor health were almost fourfold (OR = 3.88) when compared to those with a university education, and for men these odds were almost two and a half times (OR = 2.32). Material resources, in this study measured by personal income, were important factors in explaining some of the educational and ethnic differences (especially for Russian women) in poor self-rated health. Overall, we found no differences between men and women in their health ratings. On the contrary, when we controlled for physical health status, emotional distress and locus of control women reported better health than men. Health selection contributed to, but did not explain the differences by structural dimension. This study also showed a strong association of poor self-rated health with three correlates-physical health status, emotional distress and locus of control, although the influence of these correlates on poor health ratings was not seen equally in the different structural dimensions.