Metabolomics is an important "omics" technology, complementary to genomics and proteomics, as parts of systems biology, giving information (qualitative fingerprints and quantitative profiling) as a mirror of cell and extracellular metabolic activity. A cohort of small metabolites are involved in the control and regulation of cellular functions, as intermediates or final products, their presence or levels being useful for the early diagnosis of different pathologies. Bioinformatics tools are mandatory for a future "computational" metabolomics, needed to manage large number of experimentally acquired data obtained from biological samples (plants, animal or human tissues). This review presents updated information about different high-throughput analytical techniques and data acquisition software (1-2), the pre-processing of data, converted to specific matrices, further processed by specific normalization and alignment procedures (3), then analysed by statistical univariate and multivariate chemometric and /or statistical techniques (4), identifying biomarkers by comparison with databases (5), and finally elucidating the networks and pathways (6). New software is available for data conversion, pre-processing, alignment algorithms, bucketing, normalization, underlying the challenges and comparisons with international data bases. Finally, the accurate identification of individual molecules as biomarkers, either evaluated by untargeted metabolomics techniques (Principal Component Analysis -PCA), Cluster Analysis -CA) or supervised ones (Partial Least Square Discriminant Analysis (PLS-DA) is presented. The accurate identification of metabolites and their involvement in metabolic networks and pathways became possible by well-established databases (HMDB, LIPID MAPS, KEGG, etc.), to validate all experimental data. Bioinformatics is a sine-qua-non tool, to be used and valorised by untargeted or targeted metabolomics, as an integrated technology in systems biology.