Depression is a heterogeneous mental health problem affecting millions worldwide, but a majority of individuals with depression do not experience relief from initial treatments. Therefore, we need to improve our understanding of the biology of depression. Metabolomic approaches, especially untargeted ones, can suggest new hypotheses for further exploring biological mechanisms. Using the FinnTwin12 cohort, a longitudinal Finnish population-based twin cohort, with data collected in adolescence and young adulthood including 725 blood plasma samples, we investigated associations between depression and 11 low-molecular weight metabolites (amino acids and ketone bodies). In linear regression models with the metabolite (measured at age 22) as the dependent variable and depression ratings (measured at age 12, 14, 17, or 22 from multiple raters) as independent variables [adjusted first for age, sex, body mass index (BMI), and additional covariates (later)], we initially identified a significant negative association of valine with depression. Upon further analyses, valine remained significantly negatively associated with depression cross-sectionally and over time [meta-analysis beta = -13.86, 95% CI (-18.48 to -9.25)]. Analyses of the other branched-chain amino acids showed a significant negative association of leucine with depression [meta-analysis beta = -9.24, 95% CI (-14.53 to -3.95)], while no association was observed between isoleucine and depression [meta-analysis beta = -0.95, 95% CI (-6.00 to 4.11)]. These exploratory epidemiologic findings support further investigations into the role of branched-chain amino acids in depression.