Metabolite identification represents a major bottleneck in contemporary metabolomics research and a step where critical errors may occur and pass unnoticed. This is especially the case for studies employing liquid chromatography-mass spectrometry technology, where there is increased concern on the validity of the proposed identities. In the present perspective article, we describe the issue and categorize the errors into two types: identities that show poor biological plausibility and identities that do not comply with chromatographic data and thus to physicochemical properties (usually hydrophobicity/hydrophilicity) of the proposed molecule. We discuss the problem, present characteristic examples, and propose measures to improve the situation.