AIMS: Elevated lipoprotein(a) [Lp(a)] is a causal risk factor for atherosclerotic cardiovascular disease, but the mechanisms of risk are debated. Studies have found inconsistent associations between Lp(a) and measurements of atherosclerosis. We aimed to assess the relationship between Lp(a), low-density lipoprotein cholesterol (LDL-C), and coronary artery plaque severity. METHODS AND RESULTS: The study population consisted of participants of the Million Veteran Program who have undergone an invasive angiogram. The primary exposure was genetically predicted Lp(a) estimated by a polygenic score. Genetically predicted LDL-C was also assessed for comparison. The primary outcome was coronary artery plaque severity categorized as normal, non-obstructive disease, one-vessel disease, two-vessel disease, and three-vessel or left main disease. Among 18 927 adults of genetically inferred European ancestry and 4039 adults of genetically inferred African ancestry, we observed consistent associations between genetically predicted Lp(a) and obstructive coronary plaque, with effect sizes trending upward for increasingly severe categories of disease. Associations were independent of risk factors, clinically measured LDL-C and genetically predicted LDL-C. However, we did not find strong or consistent evidence for an association between genetically predicted Lp(a) and risk for non-obstructive plaque. CONCLUSION: Genetically predicted Lp(a) is positively associated with coronary plaque severity independent of LDL-C, consistent with Lp(a) promoting atherogenesis. However, the effects of Lp(a) may be greater for progression of plaque to obstructive disease than for the initial development of non-obstructive plaque. A limitation of this study is that Lp(a) was estimated using genetic markers and could not be directly assayed nor could apo(a) isoform size.