The epidemiological basis for a community approach to prevention research, programs, and policy for cardiovascular disease (CVD) is based on well-demonstrated phenomena: 1) Large population differences in cardiopulmonary disease (CPD) rates, 2) Large population differences in causal risk factors for CVD exist; 3) Population (ecologic) correlations between rates of CVD and mean levels of the established risk factors; 4) Rapid changes upward and downward in CVD death rates; 5) Rapid trends upward and downward in coronary heart disease (CHD) risk factors, principally those associated with habitual dietary patterns. 6) CHD events attributable to single and combined risk characteristics derive mainly from the center of risk factor distributions among high risk societies. Thus, a medical, high-risk approach is insufficient, and population-wide prevention strategies are needed; 7) Mean values of risk factors in youth correlate strongly with adult population distributions, and with disease rates; 8) CVD risk and all-cause mortality are significantly modified by experimental reduction of single and multiple risk factors. 9) Risky behavior can be modified in major segments of the community. Genetic susceptibility to the common modem CPD is ubiquitous. For all these reasons, research approaches and public health strategies need to be concerned with ''sick and well'' populations as well as sick and well individuals. Preventive strategies among adult high risk populations are required, but used alone, fail to influence the emergence of excess risk among low risk populations of children and youth, or among developing nations. ''Primordial prevention'' requires community strategies directed toward the prevention of risky behavior and high risk in the first place and the promotion of healthy lifestyles. (C) 1997 Elsevier Science Inc.