In order to update and improve available evidence on associations of physical activity (PA) with cardiovascular disease (CVD) by applying meta-analytic random effects modeling to data from prospective cohort studies, using high quality criteria of study selection, we searched the PubMed database from January 1980 to December 2010 for prospective cohort studies of PA and incident CVD, distinguishing occupational PA and leisure time PA, coronary heart disease (CHD) and stroke, respectively. Inclusion criteria were peer-reviewed English papers with original data, studies with large sample size (n ≥ 1,000) and substantial follow-up (≥5 years), available data on major confounders and on estimates of relative risk (RR) or hazard ratio (HR), with 95% confidence intervals (CI). We included 21 prospective studies in the overall analysis, with a sample size of more than 650,000 adults who were initially free from CVD, and with some 20,000 incident cases documented during follow-up. Among men, RR of overall CVD in the group with the high level of leisure time PA was 0.76 (95% CI 0.70–0.82, p
< 0.001), compared to the reference group with low leisure time PA, with obvious dose-response relationship. A similar effect was observed among women (RR = 0.73, 95% CI 0.68–0.78, p
< 0.001). A strong protective effect of occupational PA was observed for moderate level in both men (RR = 0.89, 95% CI 0.82–0.97, p
= 0.008) and women (RR = 0.83, 95% CI 0.67–1.03, p
= 0.089). No publication bias was observed. Our findings suggest that high level of leisure time PA and moderate level of occupational PA have a beneficial effect on cardiovascular health by reducing the overall risk of incident coronary heart disease and stroke among men and women by 20 to 30 percent and 10 to 20 percent, respectively. This evidence from high quality studies supports efforts of primary and secondary prevention of CVD in economically advanced as well as in rapidly developing countries.