Metabolomics entails identification and quantification of all metabolites within a biological system with a given physiological status; as such, it should be unbiased. A variety of techniques are used to measure the metabolite content of living systems, and results differ with the mode of data acquisition and output generation. LC-MS is one of many techniques that has been used to study the metabolomes of different organisms but, although used extensively, it does not provide a complete metabolic picture. Recent developments in technology, for example the introduction of UPLC-ESI-MS, have, however, seen LC-MS become the preferred technique for metabolomics. Here, we show that when MS settings are varied in UPLC-ESI-MS, different metabolite profiles result from the same sample. During use of a Synapt UPLC-high definition MS instrument, the collision energy was continually altered (3, 10, 20, and 30 eV) during MS acquisition. PCA and OPLS-DA analysis of the generated UPLC-MS data of metabolites extracted from elicited tobacco cells revealed different clustering and different distribution patterns. As expected, ion abundance decreases with increasing collision energy, but, more importantly, results in unique multivariate data patterns from the same samples. Our findings suggest that different collision energy settings should be investigated during MS data acquisition because these can contribute to coverage of a wider range of the metabolome by UPLC-ESI-MS and prevent biased results.