Karelia project of the Minnesota Heart Health Program. International Journal of Epidemiology 1986, 15: 176-182. Community-based cardiovascular disease control studies represent an effort to change cardiovascular disease rates in entire communities. Communities, rather than individuals, are the primary units of analysis. The cross-community multiple time series model to estimate and test the effects is based on multiple communities that are evaluated at several points over time. Issues that influence the power of the analysis include: the number of communities to be studied, community size and composition, sample sizes of surveys, the decision to use cohorts or cross-sectional surveys, the number of surveys conducted in each community, and assumptions of latencies in the effects. These points are illustrated using the experiences of the North Karelia Project and the Minnesota Heart Health Program. The North Karelia Project was a community-based cardiovascular disease (CVD) prevention programme consisting of a five-year intervention period in 1972-7. It took place in two provinces in Finland. The Minnesota Heart Health Program is similar, taking place between 1980 and 1990 in six communities in the American Midwest.