In 1977, in the evaluation of the prevention programme for cardiovascular diseases, 11,880 persons in Eastern Finland were asked to report their own weight on a questionnaire. Each participant was weighted during the following clinical examination. The data of the self-reported body weight were analysed according to sex, age, measured weight and body-mass index (BMI). The results showed that older people underestimated their weight to a greater extent than did younger people of both sexes. The error between measured and self-reported weight was greater in heavier subjects than in thinner individuals. In both sexes weight estimate error (measured weight minus self-reported weight) correlated more strongly with high BMI than with measured weight. Associations between weight estimate error and other variables were studied using a multiple regression model. Men whose annual family income was low were more likely to underestimate their weight than the men with a high annual income. In general, women reported their weight more correctly than men did. Older women were more likely to report their weight less than younger women, whereas women who visited their doctor frequently or who had higher annual family incomes were more aware of their actual body weight than those who had few doctor's consultations or whose family income was low. In men 5.2% and in women 8.3% of the variation in the weight estimate error was explained by the regression model.(ABSTRACT TRUNCATED AT 250 WORDS)