Despite its suitability to analyze polar metabolites using minute amounts of sample and its large peak capacity, CE-MS has traditionally been considered to lack the robustness required by untargeted metabolomics, especially in regulatory environments. This belief comes from the difficulty to adequately identify metabolites based on their migration times due to the variability of such parameter. In the present work, we demonstrate how this limitation can be circumvented by using standardized CE-MS conditions and automatically converting CE-MS files into electrophoretic mobility (μeff) scale. This strategy allows to conveniently exploit the advantages of CE-MS for low-volume samples generated during the toxicological risk assessment of potential neuroinflammatory substances, performed via the evaluation of astrocyte reaction. Human astrocyte cells were exposed to tumor necrosis factor alpha (TNFα) as a model compound and to digoxin at different concentrations as a tested chemical. The induced metabolic profiles were then characterized by means of CE-MS metabolomics and the success of the annotation step was evaluated and compared when one or two reference compounds were used as markers for the conversion into the effective electrophoretic mobility scale. The use of two markers resulted in a more reliable metabolite identification across all the conditions. As a result, a total of 68 anionic and cationic metabolites were annotated in both CE polarities. Unsupervised and supervised multivariate analysis enabled the comparison of the metabolomic profiles induced by each compound, highlighting common and differential metabolites, suggesting a similar but specific mechanism of activation for digoxin with regard to TNFα. CE-MS-based metabolomics is an advantageous tool for the analysis of minute amounts of samples delivered by new approach methodologies (NAMs) in chemical risk assessment, allowing high throughput toxicity screening.