The role of homocysteine as a risk factor for cardiovascular disease remains controversial because positive reports from case-control studies may be the consequence of reverse causality bias, and positive and negative results have been reported in cohort studies. This is a meta-analysis of published cohort studies. An average relative risk (ARR) was calculated using fixed and random effect models. The likelihood of publication and selection bias, and the impact of each study on the ARR were also evaluated. Fourteen eligible studies were retrieved. We found no evidence of publication bias (P =.62) nor heterogeneity (P =.56). The ARRs from a fixed effect model were 1.49 (95% CI: 1.31-1.70) for cardiac events, and 1.37 (95% CI: 0.99- 1.99) for ischemic stroke. Duration of follow up and age did not significantly change the ARRs. Hyperhomocysteinemia moderately increases the risk of a first cardiovascular event, regardless of age and follow-up duration.