We propose a strategy that integrates ultra-performance liquid chromatography/quadrupole-time-of-flight mass spectrometry (UPLC/Q-TOF-MS) and virtual docking to identify inhibitors of multiple human α-glucosidases. UPLC yielded AIB656, an acarviostatin-containing complex, which was analyzed by Q-TOF-MS to acquire structural information and was tested for inhibition of N-terminal (MGAM-N), C-terminal (MGAM-C) catalytic domain of maltase-glucoamylase, and human pancreatic α-amylase (HPA). A systematic computational study was performed to evaluate the inhibition activity for 51 identified acarviostatins with various sizes, including trace or difficult-to-prepare ingredients. We evaluated the selectivities of three α-glucosidases to acarviostatins and revealed the strong inhibition of MGAM-N by acarviostatin I0-1. The high accuracy of the dual-screening was validated using enzyme inhibition assays, and docking was suggested as a possible mechanism for the strong inhibition of MGAM-N by acarviostatin I0-1 and of MGAM-C by acarbose (acarviostatin I01). No compound in AIB656 was suitable for inhibiting all three α-glucosidases. Compared with conventional chromatographic separation and inhibitory activity detection, integrating UPLC/Q-TOF-MS identification and virtual validation was more convenient and more reliable. This strategy clearly demonstrates that MS data-based fingerprinting is a meaningful tool not only in identifying constituents in complex matrix but also in directly screening for powerful trace ingredients in natural products.