Since 1985 a small-scale community-based cardiovascular disease (CVD) preventive programme has been in operation in an inland municipality, Norsjö, in Northern Sweden. The aim of this study was to assess the development of the relationship between social position and CVD risk factors in repeated cross-sectional surveys (1985-1990) among all men and women aged 30, 40, 50 and 60 years in the study area, using an age-stratified random sample from the Northern Sweden MONICA Study of 1986 and 1990 as reference population. These multiple cross-sectional surveys comprised a self-administered questionnaire and a health examination. Of the study population 95% (n = 1499) and 80% of those in the reference area (n = 3208) participated. Subjects were classified with regard to demographic, structural and social characteristics in relation to CVD risk factors and self-reported health status. Time trends in classical risk factor occurrence were assessed in terms of age- and sex- adjusted odds ratios using Mantel-Haenszel procedures. When simultaneously adjusting for several potential confounders we used a logistic regression analysis. Initially, more than half of the study population, both males and females, had and elevated (> or = 6.5 mmol/l) serum cholesterol level. After adjustments had been made for age and social factors it was found that the relative risk of hypercholesterolaemia dropped substantially and significantly among both sexes during the 6 years of CVD intervention in the study area. However, the probability of being a smoker was significantly reduced only in highly educated groups. Among other risk factors no single statistically significant change over time could be found. In the reference area there were no changes over time for the selected CVD risk factors. People in the study area had a less favourable perception of their health than those in the reference area. Social differences were found when perceived good health was measured, especially in variables indicating emotional and social support. When sex, age and social factors had been accounted for there was not clear change over the years in perceived good health.