The authors studied the effectiveness of community-wide health education on physical activity knowledge, attitudes, self-efficacy, and behavior. Random samples of residents aged 18-74 years who lived in four central California cities (baseline, n = 1,056 men and 1,183 women) were evaluated in 1979-1980 and approximately every 2 years thereafter to obtain four independent samples. Moreover, every subject in the initial independent samples was asked to return for follow-up every 2 years thereafter; subjects who completed all four examinations constituted the cohort sample (n = 408 men and 499 women). Two medium-sized cities received health education and two similarly sized cities served as controls. Results indicated little consistent evidence of a treatment effect on physical activity knowledge, attitudes, or self-efficacy in either men or women. Among physical activity measures, there was an indication of a positive treatment effect for men in the independent samples for estimated daily energy expenditure and percent participation in vigorous activities (p < 0.01), and for women in the independent (p = 0.014) and cohort (p < 0.01) samples for engagement in the number of moderate activities. These results underscore the need for development of more effective interventions to change physical activity than is provided by a broad-based, community-wide health education program and for more sensitive and reliable measures of knowledge, attitudes, and behavior with regard to physical activity.