The aim of the present study was to determine whether lipoprotein(a) [Lp(a)], considered a causal risk factor for cardiovascular disease, primarily promotes thrombosis or atherosclerosis. Using a Mendelian randomization study design, we measured plasma Lp(a) and genetically elevated Lp(a) levels through the LPA kringle IV type 2 repeat genotype in 41231 individuals. We included 2 general population studies of both venous thrombosis and combined thrombosis and atherosclerosis in coronary arteries (=myocardial infarction), and 3 -case--control studies of atherosclerotic stenosis. Neither Lp(a) tertiles nor LPA kringle IV type 2 tertiles associated with the risk of venous thrombosis in general population studies (trend: P=0.12-0.76), but did each associate with risk of coronary, carotid, and femoral atherosclerotic stenosis in -case--control studies (trend: P<0.001 to 0.04). Lp(a) and LPA kringle IV type 2 tertiles also associated with the risk of myocardial infarction in general population studies (trend: P<0.001 to 0.003). For doubling of Lp(a) levels, instrumental variable estimates of hazard/odds ratios were 1.02 (95% CI 0.90-1.15) and 1.04 (0.93-1.16) for venous thrombosis in the 2 general population studies, 1.12 (1.01-1.25), 1.17 (1.05-1.32), and 1.16 (1.01-1.35), respectively, for coronary, carotid, and femoral atherosclerotic stenosis in -case-control studies, and 1.21 (1.10-1.33) and 1.17 (1.05-1.29) for myocardial infarction in general population studies. This supports that Lp(a) primarily promotes atherosclerotic stenosis rather than venous thrombosis.