Molecular phylogenetics based on nucleotide sequence comparisons has profoundly influenced plant taxonomy. A comprehensive chemotaxonomical approach based on GC-MS and UHPLC-HRMS profiling was evaluated for its ability to characterize a large collection of plants all in the violet family Violaceae (n = 111) and thus decipher the taxonomy. A thorough identification of violets is challenging due to their natural hybridization and phenotypic variability. Phylogenetic inference performed on ribosomal internal transcribed spacer sequences using maximum likelihood and neighbor-joining distance methods allowed the clear identification of 58% of the collection. Metabolomic approaches with multivariate data analysis were performed on SPME/GC-MS chromatograms of volatile compounds emitted by fresh mature flowers and on UHPLC-HRMS/MS leaf extracts for non-volatile compounds. Interestingly, molecular and biochemical approaches provided separate classifications while highlighting several common clusters. The profiling of secondary metabolites was proved most suitable for the classification of hundreds of extracts. The combination of phylogenetic and chemotaxonomic approaches, allowed the classification of 96% of the entire collection. A correlation network revealed specific chemotaxonomic biomarkers, in particular flavonoids, coumarins and cyclotides. Overall, our pioneering approach could be useful to solve misclassification issues within collections of close plant species.