A re-examination after 3 years was done in 1975 in a 20% random subsample (n = 1683) of the representative population sample (males and females, 25-59 years) that was examined in 1972 in North Karelia (NK), and a matched reference county as the baseline survey for the community programme in NK. The changes in smoking habits, serum cholesterol, dietary fat consumption and systolic BP were more favourable among the subjects in the NK sample than among the reference sample, although the differences were generally small. Results from multivariable analyses are presented to show the variables that predict a favourable risk factor change in the individual. Living in NK is associated in the analysis with a favourable change in each of the three risk factors. The limitation of this method in the evaluation of a community programme is discussed.