BACKGROUND AND PURPOSE: The relationship between elevated lipoprotein (a) levels[Lp(a)] and stroke is controversial. We systematically reviewed the literature to determine whether Lp(a) is a risk factor for stroke. METHODS: We searched MEDLINE (1966 to 2006), EMBASE (1974 to 2006), and Google scholar for articles on Lp(a) and cerebrovascular disease. From potentially relevant references retrieved, we excluded uncontrolled studies, studies of children with stroke, studies investigating carotid atherosclerosis, and studies lacking adequate data. RESULTS: Thirty-one studies comprising 56 010 subjects with >4609 stroke events met all inclusion criteria and were included in the meta-analysis. In case-control studies (n=23 with 2600 strokes) unadjusted mean Lp(a) was higher in stroke patients (standardized mean difference, 0.39; 95% CI, 0.23 to 0.54) and was more frequently abnormally elevated (OR, 2.39; 95% CI, 1.57 to 3.63). Sensitivity analysis and meta-regression did not find any influence of study design, measurement period of Lp(a) in relationship to stroke episode, subtype, age, and sex to explain the substantial heterogeneity between studies (I(2)=83.7%; P<0.001). There was no evidence of publication bias. In nested case-control studies (n=3 with 364 strokes) Lp(a) was not a risk factor for incident stroke (OR, 1.04; 95% CI, 0.6 to 1.8). In prospective cohort studies (n=5 with >1645 strokes), incident stroke was more frequent in patients in the highest tertile of Lp(a) distribution compared with the lowest tertile of Lp(a) (RR, 1.22; 95% CI, 1.04 to 1.43). There was no publication bias or heterogeneity in the prospective studies (I(2)=0.00%; P=0.67). CONCLUSIONS: This meta-analysis suggests that elevated Lp(a) is a risk factor for incident stroke.