STUDY OBJECTIVES: Lipoprotein (a) [Lp(a)] is a biomarker of atherosclerotic cardiovascular disease, but its role in mental disorders is controversial. Our study aimed to explore the causality between Lp(a) levels and mental disorders by combining retrospective and Mendelian randomization (MR) studies. METHODS: All genome-wide association study datasets used in the MR study were obtained from UK Biobank, FinnGen, and the Psychiatric Genomics Consortium. The matched case-control study were based on electronic health records from the Second Affiliated Hospital of Nanchang University and NHANES III cohort. RESULTS: In the MR analysis, Lp(a) had a positive causal effect on the longest period of depression [1.05 (1.02-1.08), P = 0.0001], the number of depressive episodes [1.03 (1.01-1.06), P = 0.009] and a weak negative effect on memory loss [0.84 (0.72-0.99), P = 0.039]. Meanwhile, bipolar and major depressive disorder status was causally associated with significantly lower Lp(a) levels [0.96 (0.93-0.98), P = 0.003]. Retrospective study revealed low Lp(a) levels were associated with a significantly higher risk of depression (n = 670) [1.273 (1.007, 1.609), P = 0.044], anxiety (n = 1284) [1.231 (1.041, 1.456), P = 0.015] and major depression (n = 538) [1.364 (1.012,1.841), P = 0.042]. CONCLUSIONS: This study found there is a causal relationship between the number and longest period of depressive episodes or memory loss and Lp(a), while bipolar disorder and major depressive disorder were associated with a significant causal effect on reduced Lp(a) levels. Future studies should focus on whether a sustained decrease in Lp(a) levels could cause the development of mental disorders, and which target value is suitable for clinical practice.