Direct infusion-mass spectrometry (DI-MS) currently serves as an alternative analytical tool for metabolomics owing to the unique high-throughput advantage. Except the inherent shortcoming at a significant matrix effect, there are two other primary technical obstacles dampening its wide applications, such as data alignment and structural annotation. To address these two obstacles, a novel strategy termed as DI-three-dimensional-MS (DI-3D-MS) was proposed here, and chemome comparison among several confusing herbal medicines (HMs) belonging to the Umbelliferae family was conducted as a proof-of-concept. Each test sample was directly infused into Qtrap-MS. In the first dimension, stepwise multiple ion monitoring (MIM) program was implemented to universally acquire the quantitative information on all HMs and to generate aligned data files. In the second dimension, MS2 spectra were universally recorded by enhanced product ion (EPI) experiments that were triggered by MIM via an information-dependent acquisition algorithm. In the third dimension, online energy-resolved MS (ER-MS) was programmed to yield breakdown graphs for all MIM items. Moreover, a data library was built to aid structural identification by involving MS2 and CE50 features that were obtained by well-developed LC-MS methods. Qualitative and quantitative potentials of DI-3D-MS were validated toward metabolomics study. Significant species differences were observed, and all materials were grouped into three clusters. After matching MS2 spectra and breakdown graphs between DI-3D-MS and those in the data library, coumarins ubiquitously existed in each HM, and angular-type pyranocoumarins, linear-type pyranocoumarins, angular-type furanocoumarins, along with ligustilide derivatives offered primary contributions for the classification pattern. Above all, DI-3D-MS is an eligible choice for rapid metabolomics of HMs and other matrices as well.