Non-alcoholic fatty liver disease (NAFLD) and depression are common disorders and have bidirectional contributing relationships to metabolic syndrome. We aimed to determine whether a fasting serum signature of recent, self-reported depressive symptoms could be identified in a heterogeneous NAFLD cohort using nuclear magnetic resonance (NMR)-based metabolomics integrated with clinical chemistry. Serum nuclear magnetic resonance (NMR) metabolite profiles and corresponding clinical chemistry were compared between patients with depressive symptoms in the last 12-months (n = 81) and patients without recent depressive symptoms (n = 137 controls) using multivariate statistics. Orthogonal partial least squares discriminant analysis (OPLS-DA) of the biochemical and metabolomic data identified NAFLD patients with recent depression with a cross-validated accuracy of 61.5%, independent of age, sex, medication, and other comorbidities. This led to the development of a diagnostic algorithm with AUC 0.83 for future testing in larger clinical cohorts. Serum triglycerides, VLDL cholesterol, and the inflammatory biomarker GlycA were key metabolites increased in patients with recent depressive symptoms, while serum glutamine level was reduced. Here, serum NMR metabolite analysis provides a link between disturbed lipid metabolism, inflammation, and active mental health issues in NAFLD, irrespective of disease severity.