Metabolomics can be defined as the scientific field aiming at characterizing all low-weight molecules (so-called metabolites) in a biological system. At the time of death, the level and type of metabolites present will most likely reflect the events leading up to death. In this proof of concept study, we investigated the potential of post-mortem metabolomics by identifying post-mortem biomarkers, correlated these identified biomarkers with those reported in clinical metabolomics studies, and finally validated the models predictability of unknown autopsy cases. In this post-mortem metabolomics setting, ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry data from 404 post-mortem samples, including pneumonia cases and control cases, were processed using XCMS (R). Potential biomarkers were evaluated using principal component analysis and orthogonal partial least squares-discriminant analysis. Biomarkers were putatively annotated using an in-house database and the online databases METLIN and HMDB. The results showed that clear group separation was observed between pneumonia cases and control cases. The metabolites responsible for group separation belonged to a broad set of biological classes, such as amino acids, carnitines, lipids, nicotinamides, nucleotides, and steroids. Many of these metabolites have been reported as important in clinical manifestation of pneumonia. For the unknown autopsy cases, the sensitivity and specificity were 86 and 84%, respectively. This study successfully investigated the robustness and usability of post-mortem metabolomics in death investigations. The identified post-mortem biomarkers correlated well with biomarkers reported and identified through clinical research.